| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Natural Resources Research Institute, Univ. of Minnesota, 5013 Miller Trunk Highway, Duluth, MN 55811
b Dep. of Biological Sciences, Univ. of Notre Dame, Notre Dame, IN 46556
c National Center for Environmental Assessment, U.S. Environmental Protection Agency, 26 W. Martin Luther King Drive, Cincinnati, OH 45268
Corresponding author (cjohnsto{at}d.umn.edu)
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
6 nmol N2O g-1 h-1) and in organic backwater zones (8.3 nmol N2O g-1 h-1 at Fond du Lac, 4.8 nmol N2O g-1 h-1 at Pokegama), but denitrification was consistently NO-3-limited throughout both wetlands. Riverbeds were zones of highest P concentration in soil, vegetation, and summer surface water. Sedimentation rates were higher in riverbeds (289 g m-2 d-1 at Fond du Lac, 54 g m-2 d-1 at Pokegama) than in backwaters (80 g m-2 d-1 at Fond du Lac, 17 g m-2 d-1 at Pokegama). The two backwater zones had comparably low summer surface water concentrations of NO3–N (4 µg L-1), NH4–N (6 µg L-1), total P (TP) (80 µg L-1), total suspended solids (TSS) (6 mg L-1), and volatile suspended solids (VSS) (4 mg L-1). This seasonal convergence of surface water chemistry implies that biotic processes common to the two backwater areas override their substrate differences. Backwaters were hydrologically connected to the river mainstem via openings in discontinuous natural levees, but the different water chemistry of riverbed vs. backwater zones indicated minimal water exchange between them. This hydrologic zonation of riverine wetlands by geomorphic structures was the major source of intra-wetland variability.
Abbreviations: AFP, acid-fluoride extractable P • SRP, soluble reactive P • TN, total N • TP, total P • TSS, total suspended solids • VSS, volatile suspended solids
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Riverine wetlands play an important role in regulating fluxes of waterborne material (Wetzel, 1990; Johnston, 1993). Mechanisms by which wetlands remove nutrients from river waters include sedimentation, organic matter accumulation, chemical sorption, denitrification, and biotic uptake (Johnston et al., 1997). Geomorphology can potentially influence these mechanisms by altering flooding frequency and duration, hydrologic connectivity with the river channel, soil texture, soil organic matter content, soil aeration, and plant growth. Wetland classification schemes are frequently based on geomorphology because of the strength of the interaction between geomorphic structure and ecological function (Brinson, 1993).
Sedimentation rates vary among geomorphic landforms in riverine wetlands (reviewed by Johnston et al., 1997). In some riverine wetlands, backwater areas exhibit higher sedimentation rates than streamside locations (Kuenzler et al., 1980; Baumann et al., 1984; Kleiss, 1996), whereas in others the relationship is reversed (DeLaune et al., 1978; Baumann et al., 1984; Johnston et al., 1984a; Kadlec and Robbins, 1984) or shows no consistent spatial pattern (Mitsch et al., 1979). Sedimentation is an effective means of nutrient retention, sequestering 1 to 13 g N m-2 yr-1 and 0.1 to 2.6 g P m-2 yr-1 (reviewed by Johnston, 1991).
Soil texture varies substantially across river floodplains, predictably affected by differences in flow velocity and turbulence (Hjulstrøm, 1939; James, 1985). The organic matter content of floodplain soils also typically increases with distance from actively flooded stream banks to less actively flooded slackwater areas (Johnston et al., 1997). Differences in soil nutrient content accompany variation in soil texture and organic matter, with TP inversely related to grain size, and total N (TN) increasing with organic matter content (Johnston et al., 1984b, 1997). Fine-textured and highly organic soils also tend to contain more amorphous Fe and Al compounds, which sorb P (Richardson, 1985; Sah et al., 1989; Lockaby and Walbridge, 1998). Organic soils also have high cation-exchange capacities (Puustjärvi, 1956), which can increase NH+4 sorption (Burge and Broadbent, 1961; DeBusk and Reddy, 1987).
Subtle topographic variations in riverine wetlands affect N forms and availability, because aerobic conditions are required for nitrification, whereas anaerobic conditions are required for denitrification or dissimilatory NO-3 reduction. The slight elevation of natural levee soils causes them to be better aerated than soils of perennially flooded geomorphic features (DeLaune et al., 1983), resulting in high concentrations of soil NO-3 due to nitrification (Johnston, 1993). Denitrification is the dominant mechanism of N removal in most riparian soils (Seitzinger, 1988, 1994; Pinay et al., 1993; Hanson et al., 1994; Hill, 1996), but dissimilatory NO-3 reduction can occur in soils that are highly reduced and have low C contents (Buresh and Patrick, 1978). Denitrification is often NO-3 limited, and is occasionally C limited (Tiedje et al., 1982). The highest denitrification rates occur where the close proximity of aerobic and anaerobic zones minimizes the path length for diffusion, or where aerobic and anaerobic conditions alternate over time (Reddy et al., 1976; Smith and Patrick, 1983).
Vegetation biomass and net primary productivity are generally highest in streamside levees that are slightly elevated but receive frequent flood pulses (Conner and Day, 1976; DeLaune et al., 1983). Mechanisms that control plant productivity include soil fertility (Day et al., 1988) and porewater S-, which influences plant NH+4 uptake (Bradley and Morris, 1990). However, plants are a minor long-term nutrient removal mechanism in wetlands, except in young aggrading forested wetlands, because the majority of the nutrients taken up during the course of a growing season are leached out of the green tissues into wetland surface waters or returned to the wetland soil surface in litterfall (Klopatek, 1978; Johnston, 1991; Lockaby and Walbridge, 1998).
We examined the seasonal influence of soils and geomorphology on nutrient forms and concentrations in riverine wetlands in northeastern Minnesota (silty soils) and northwestern Wisconsin (clayey soils). Soil, water, and plant biogeochemistry were contrasted between and within the wetlands according to geomorphic features (riverbed, levee, and backwater zones). Specific hypotheses were:
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
Transects of sample points, oriented perpendicular to the long axis of flow in the rivers and spaced 20 m apart, were established in each of the wetlands. Each of the fifteen transects at the Pokegama wetland consisted of a sample point in the riverbed, on the levee, and at two to five locations in the backwater. Each of the nine transects at the Fond du Lac wetland consisted of sample points in the deep riverbed (water depth 45 cm), in the shallow riverbed (water depth 25 cm), on the levee, and at three to six locations in the backwater. Points were spaced 20 m apart along the transects in backwater areas, but were closer (4–10 m) in the riverbed and on the levee, where environmental gradients changed rapidly. The total number of sample points was 75 at Fond du Lac and 83 at Pokegama.
The study was initiated during the fall (October 1992) at the Pokegama site and during the spring (May 1993) at the Fond du Lac site; soil and water samples were collected during the spring and summer (August 1993) at both wetland sites. Tables 1 to 6 and Fig. 2 to 8 summarize the results of 56 variables and 8343 individual measurements.
|
|
|
|
|
|
|
|
|
|
|
|
|
Soil moisture and bulk density were measured for every soil core so that soil nutrient concentrations could be expressed by dry weight and by volume. Bulk density was determined by measuring the wet weight of each core and converting to dry-mass equivalent per unit volume of soil using moisture data and the known volume of the cores (Klute, 1986). Soil moisture was determined gravimetrically on two subsamples from each core that were dried at 105°C. A subsample of the field-moist core was centrifuged to remove porewater, from which an aliquot was frozen for later analysis (see below).
Particle-size distribution was determined by the pipette method following digestion with 30% H2O2 and dispersion with sodium hexametaphosphate (Gee and Bauder, 1986). Soil organic matter content was determined by weight loss on ignition in a muffle furnace at 550°C. Amorphous Fe and Al were analyzed using an atomic absorption spectrophotometer (Varian Spectra-30A, Varian Instruments, Sugar Land, TX) following acid-ammonium-oxalate extraction (Jackson et al., 1986). Soil pH and S- concentrations were determined in the laboratory on a 1:1 soil/deionized water slurry using a pH–ion-specific meter (Model 290A, Orion Research, Beverly, MA). Acid-fluoride extractable P (Olsen and Sommers, 1982) was measured with a QuickChem 4 autoanalyzer (Lachat Instruments, Milwaukee, WI). Following soil extraction with 2 M KCl (Keeney and Nelson, 1982), a QuickChem 4 autoanalyzer was used to measure NH4–N (salicylate method) and NO3–N plus NO2–N (sulfanilamide method; Owen and Axler, 1992).
Total macronutrient (C, N, P) analyses were performed at the U.S. Geological Survey, Laurel, MD. Total C and TN were determined using an elemental analyzer (Model CE440, Leeman Labs, Hudson, NH), using acetanilide as an internal standard for the machine, and Bureau of Standard pine and wheat flour samples as external standards. Total P was determined by digestion of dried soil with 50% H2SO4 at 300°C for 1 h and analysis by inductively coupled plasma spectrometry (Optima 3000, Perkin Elmer, Norwalk, CT).
All soil nutrient concentrations were multiplied by bulk density (dry mass per unit volume of soil) to convert to per volume units. This was done because (i) it allowed both soil and water concentrations to be expressed in volumetric units, facilitating their comparison; (ii) it permitted more realistic comparison of nutrient concentrations in organic and mineral soil samples, which had bulk density values spanning an order of magnitude within the wetlands studied; and (iii) soil roots occupy a volume of soil rather than a mass of soil, so volumetric units are more realistic in terms of plant nutrient availability (Mehlich, 1972).
Sedimentation rates were measured at inundated sample points in riverbed and backwater zones during 1993 and 1994 using sediment trap bottles anchored to the bottom (Gardner, 1980). Each 500-mL Nalgene bottle was filled with water and frozen prior to installation to minimize contamination with sediments disturbed by the installation process (Fennessy et al., 1994). In 1993, sediment trap bottles were installed in late June and retrieved in mid October; in 1994, sediment trap bottles were installed in mid May and retrieved in mid October. Bottle contents were dried, and total mass accumulated per unit area (bottle mouth opening = 15.13 cm2) was divided by the number of days in place to compute sedimentation rates. An unanticipated result was that many of the bottles were used as refugia by large macrofauna (bullheads [Ameiurus spp.], crawfish [Orconectes spp.]) in 1993, and those samples were discarded. The sample design was modified in 1994 by tying 1-cm mesh nylon netting over the bottle mouth to exclude large aquatic organisms.
Denitrification rates were determined using the acetylene block technique (Balderston et al., 1976; Yoshinari et al., 1977; Knowles, 1982, 1990). Aliquots (8 cm3) of homogenized soils collected from each sample point in spring and summer were dispensed into duplicate, preweighed 60-mL serum vials, and reweighed. The samples were slurried with an equal volume (8 mL) of autoclaved deionized water. The vials were then crimp-sealed, sparged for 5 min with oxygen-free N, and reweighed. After sparging, 10 cm3 of high purity analytical-grade acetylene was added to the vials, and a 2.5-cm3 (time zero) sample was withdrawn using a gas-tight syringe and injected into an evacuated 5-mL Vacutainer (Becton-Dickinson Vacutainer Systems, Rutherford, NJ). The Vacutainers were frozen until analysis. After sampling, a small amount of silicone vacuum grease was applied to the top of the serum stopper, and the vials were placed on an orbital shaker table and incubated in the dark at field temperatures (15°C in spring, 20°C in summer). The vials were resampled the following day, and subsequently every 2 d throughout the incubation. After 6 d of incubation, one set of replicates was amended with 1 mL of autoclaved oxygen-free KNO3 solution (final concentration of 100 µmol). Killed controls (using formalin) and amendment controls (addition of 1 mL of autoclaved oxygen-free deionized water) were also prepared and sampled. Denitrification was measured as N2O produced during the course of the incubation. Nitrous oxide was quantified on a gas chromatograph (Model 14A, Shimadzu Instruments, Columbia, MD) equipped with an electron capture detector and fitted with a 3-m Porapak-Q column (Alltech Associates, Deerfield, IL) under isothermal conditions (detector 300°C, injector 150°C, column 60°C), using oxygen-free N at 30 mL min-1 as the carrier gas. Denitrification rates were calculated from the linear portion of N2O accumulation curves. At the conclusion of the experiment, vial headspace volumes were determined by filling the serum vials with water and recording their weight. Volume-specific soil wet and dry weights were used to normalize rates.
Water Sampling and Analysis
Water depth was measured at each inundated sample point during August 1993 using a meter stick attached to a 41 by 41 cm plywood base designed to rest on the soil surface without penetrating it. Two sets of surface water samples were collected during May and August 1993 at each inundated sample point, transported on ice to the laboratory, and stored at 4°C or frozen prior to analysis. The first sample set was used to determine water pH within 48 h (Orion model 290A pH meter, Orion Research, Boston, MA), after which the water was filtered (Whatman GF/C glass-fiber filter) to determine TSS (residue weight after drying at 105°C) and VSS (residue weight after ignition at 550°C). The second sample set was divided into filtered (0.45-µm filter) and unfiltered portions, and frozen until analysis. Unfiltered samples were digested for TN and TP (persulfate–NaOH simultaneous digestion method; Owen and Axler, 1992). Phosphorus was measured in a spectrophotometer by the manual ascorbic-acid method (Murphy and Riley, 1962), using digested surface water samples to determine TP and using filtered surface water and soil porewater samples to determine soluble reactive P. A QuickChem 4 autoanalyzer was used to measure NH4–N (salicylate method) and NO3–N (sulfanilamide method) in filtered surface and porewater, and TN in the digested surface water sample. All water analyses follow standard analytical chemistry and quality assurance protocols, described in detail by Owen and Axler (1992).
Vegetation Sampling and Analysis
Peak aboveground herbaceous biomass of both emergent and submergent vegetation was measured in August 1993 by clipping all vegetation at the sediment surface within a 0.25-m area adjacent to each sample point. Samples were air-dried and weighed. Moisture content was determined on subsamples by drying in a 60°C oven for 48 h and was used to convert the air-dried weights to oven-dried. Biomass of woody species growing on the levees was not determined. Dried vegetation subsamples were ground with a Udy grinder (Udy Corp., Fort Collins, CO). Total C, N, and P concentrations in vegetation were determined as described above for total soil macronutrients.
Statistical Analyses
Samples were classified by the geomorphic zone from which they were collected: riverbed, levee, backwater mineral, and backwater organic. Riverbed, levee, and backwater were distinguished in the field on the basis of their physical arrangement relative to the river. Backwater organic samples were those backwater sample points containing >120 g kg-1 organic C in surface soil samples (0–15 cm). Although soils at some of the backwater organic sample points appeared to be Histosols (Soil Survey Staff, 1996), it was not possible to classify the sample points based on soil morphology because of their constant inundation. Therefore, soils in riverbed, levee, and backwater mineral zones are clearly mineral soils, whereas soils in backwater organic zones may or may not be Histosols.
The Box–Cox method (Box and Cox, 1964) was used to determine the best transformation (usually the natural logarithm) to normalize the data for analysis of variance (transformations listed in Tables 1, 3, 5, and 6). Bartlett's test was used to examine the assumption of homogeneity of variance, and the Wilke-Shapiro test was used to verify that the residuals had a normal distribution. Two-way analysis of variance was used to analyze the effects of geomorphology, wetland site (Fond du Lac vs. Pokegama), and their interaction. When the main effect of geomorphology was significant at the P < 0.05 level, F tests were used to compare pairs of geomorphic zones (reported in Geomorphology Contrasts columns of Tables 1, 3, 5, and 6). When there was significant interaction between geomorphology and wetland site, a new variable was created that combined the levels of the main effect, and a second analysis of variance was performed, using the Tukey–Kramer multiple comparison of means procedure to compare groups (reported in Geomorphic Relationships columns of Tables 1, 3, 5, and 6). Seasonal differences among soil, water, and porewater chemical characteristics were determined by two-way analysis of variance with repeated measures using pooled spring and summer data from both wetlands, and with pooled spring, summer, and fall data from the Pokegama wetland alone (Table 2). A threshold of P < 0.05 was used to detect significant differences. All statistical analyses were done with PC SAS (SAS Institute, 1989).
In this manuscript, the term site refers to the two wetlands (Fond du Lac and Pokegama), zone refers to geomorphic zone (riverbed, levee, mineral backwater, organic backwater), and sample point refers to an individual sample location.
|
RESULTS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
However, the two wetlands had very different soil textures (significant effect of site, Table 1). Soils in the Pokegama wetland contained substantially more clay and less silt than those in the Fond du Lac wetland (Fig. 2c and 2d). This difference in soil texture is due to differences in parent material between the two sites: soils in the Pokegama wetland developed from clayey glacio-lacustrine deposits, whereas those in the Fond du Lac wetland are derived from silty alluvium (Clayton, 1984). Soils in the Fond du Lac wetland also contained less TN but more organic matter, TP, and Fe than did soils in the Pokegama wetland (Fig. 2e, 3b–3d). Plants in the Fond du Lac wetland contained less C and more N than did those at Pokegama (Fig. 4b and 4c).
Effect of Within-Wetland Geomorphology
Much of the variation within the wetlands was explained by dividing the data into four geomorphic zones: riverbed (R), levee (L), backwater mineral (M), and backwater organic (O). There were significant differences among geomorphic zones for water depth, plant N and P concentrations, and all soil characteristics except Fe and Al concentrations (Table 1, Fig. 2–4). Averages for most of the variables increased or decreased along the gradient R-L-M-O: increasing clay, soil organic matter, soil moisture, soil C, and soil N (Fig. 2d, 2e, 2f; 3a, and 3b); decreasing sand, bulk density, and soil P (Fig. 2b, 2g, and 3c). For other variables, individual geomorphic features were significantly different than others. For example, riverbed soils contained less silt (Fig. 2c) and riverbed plants contained more P and less N than plants growing elsewhere in the wetlands (Fig. 4c and 4d). The shallowest water depth occurred at the levees (Fig. 2a).
For Fe and Al, there was a significant interaction between site and geomorphic features, despite no significant main effect of either, indicating that the effect of geomorphic features on Fe and Al concentrations was different at the two wetlands (Table 1). For example, soils in levee and organic zones had the highest Fe concentrations at Fond du Lac, but the lowest Fe concentrations at Pokegama; soils in riverbed and backwater mineral zones had comparable Fe concentrations at both sites (Fig. 3d). For both Fe and Al there was considerable overlap among Tukey groupings (Geomorphic Relationships, Table 1).
Soil Inorganic Nutrients and pH
Seasonal Effects
With the exception of NH4–N, there were significant concentration differences between spring and summer for every chemical form measured (Table 2, Fig. 5 and 6). Sulfide and porewater NO3–N concentrations decreased between spring and summer, whereas all other parameters increased. The lower summer S- concentrations (Fig. 5b) are an indication that soils were more aerobic in summer than in spring, probably due to increased photosynthesis. Although it seems contradictory that NO3–N concentrations increased in soil but decreased in porewater between spring and summer, the seasonal trend for porewater was dominated by the extremely high NO3–N concentrations in the levees, which were higher in spring than in summer (Fig. 6b and 6e). The spring to summer increase in soil pH occurred primarily in the Fond du Lac riverbed and levee (Fig. 5a); there was little change elsewhere in the two wetlands. Ammonium-N concentrations did not change significantly between spring and summer in porewater or soil.
Nutrient concentrations were higher in fall than in spring or summer for most forms in porewater and surface water (Table 2). Porewater NO3–N concentrations were nearly an order of magnitude higher in fall than in summer, whereas porewater soluble reactive P (SRP) concentrations were significantly lower in fall than in summer (Fig. 6f). Fall soil NH4–N concentrations were significantly lower than those in spring or summer (Fig. 5h), although porewater NH4–N concentrations were about four times higher in fall than in spring or summer (Fig. 6d). There was no significant seasonal difference in soil pH when the Pokegama site was considered alone (Fig. 5f).
Given the many seasonal effects observed with the Pokegama and combined data sets, subsequent analyses were done separately for spring and summer.
Comparisons between Wetlands
In the spring, inorganic nutrient concentrations in both bulk soil and porewater differed between the two wetlands for almost every parameter measured (Table 3). Relative to the Pokegama wetland, the Fond du Lac wetland had higher soil moisture, lower soil pH, higher S-, lower soil NO3–N concentrations, higher soil and porewater NH4–N concentrations, and higher soil acid-fluoride extractable P (AFP) but lower porewater SRP concentrations (Fig. 5 and 6). The two wetlands had comparable porewater NO3–N concentrations, regardless of season.
In the summer, there were fewer differences between the two wetlands. As in spring, the Fond du Lac wetland had lower soil pH, higher S-, and higher concentrations of porewater NH4–N and soil AFP than did the Pokegama wetland, but other inorganic nutrient concentrations (soil NH4–N, soil and porewater NO3–N, and porewater SRP) were comparable.
Within-Wetland Spatial Variability
Geomorphology affected soil pH and most inorganic nutrients in both seasons (Table 3). In spring, all parameters were significantly affected by geomorphology. In summer, geomorphology had no significant effect on porewater and soil NH4–N and soil NO3–N when Fond du Lac and Pokegama data were considered together (Table 3), but it did significantly affect these inorganic N forms (P < 0.0005) when the Pokegama data were considered alone. Concentrations of NH4–N were lowest in the Pokegama levee, highest in organic backwater, and intermediate in riverbed and mineral backwater zones (Fig. 5h and 6d). Concentrations of soil NO3–N were highest in the riverbed and levee, and lowest in the backwater (Fig. 5i).
The influence of soil aeration was evident in S- concentrations, which were highest in backwaters and lowest in levees in both spring and summer. The more aerobic conditions in the levees were also related to the pattern of NO3–N distribution, which was highest in levee porewater and soils. The nitrification of NH4–N to NO3–N, a process requiring O2, could explain the high NO3–N concentrations in the levees. This mechanism could also explain the depleted concentrations of soil and porewater NH4–N at Pokegama. Spring NH4–N concentrations at both sites were highest in backwater organic zones, where the mineralization of organic N in the absence of nitrification would cause NH4–N to accumulate (Fig. 5c, 5h, 6a, and 6d).
The Pokegama levee was more elevated than the Fond du Lac levee, a difference that altered their seasonal trends for soil and porewater inorganic N. A summer flood that submerged two-thirds of the Fond du Lac levee sample points resulted in increased NH4–N concentrations, even though NH4–N concentrations at Pokegama and elsewhere at Fond du Lac decreased or remained constant between spring and summer (Fig. 5c and 6a). Summer soil NO3–N concentrations were also low in the Fond du Lac levee, in contrast to the elevated NO3–N concentrations in Pokegama levee soils (Fig. 5d and 5i). Dissimilatory reduction of NO3–N to NH4–N could explain this spring to summer increase in NH4–N and decrease in NO3–N in the partially flooded Fond du Lac levee.
In the spring, soil pH was highest in the levee and riverbed at Pokegama, lowest in the levee and riverbed at Fond du Lac, and intermediate in the backwater areas of both wetlands (Fig. 5a and 5f). By the summer, the pH of soils in the Fond du Lac riverbed had increased by 0.7 pH units to an average value of 6.8, and the pH of soils in the Fond du Lac levee had increased by 0.4 pH units to an average value of 6.7 (Fig. 5a). Seasonal changes in soil pH in the Fond du Lac backwater and throughout Pokegama were much smaller (0.2 pH units).
Summer porewater SRP concentrations were highest in riverbed and levee soils and lowest in backwater soils for both sites; spring porewater concentrations were also significantly higher in levees than in backwater mineral soils (Fig. 6c and 6f). Geomorphology significantly affected soil AFP concentrations in both seasons, mainly due to significantly lower concentrations in the Pokegama levee; other geomorphic zones were poorly separated (Fig. 5e and 5j).
Surface Water Inorganic Nutrients and pH
Standing water covered the soil surface in the riverbed and most of the backwater area of both wetlands during the entire duration of the study (Fig. 2a). Surface waters in the two wetlands contained very low concentrations of nutrients, and all concentrations were expressed in mircrograms per liter. Most of the N in surface water was in organic form. Inorganic N forms (Fig. 7a, 7b, 7g, and 7h) constituted <10% of TN (Fig. 7d and 7j). The ratio of soluble reactive P (Fig. 7c and 7i) to TP (Fig. 7e and 7k) ranged from 10 to 25%.
Seasonal Effects
The pooled spring and summer data from both wetlands showed there were significant seasonal effects for every water parameter (Table 2). Consistent with the seasonal trends observed for inorganic soil nutrients, concentrations of NO3–N and SRP were higher in the summer. Average concentrations of TN and TP were also higher in the summer, although there was no seasonal difference in TP concentrations when the Pokegama wetland was considered alone (Table 2, Fig. 7d, 7e, 7j, and 7k). Water pH was higher in spring than in summer (Fig. 7f and 7l), the opposite of the seasonal trend for soil pH, as was NH4–N (Fig. 7a and 7g). In the Pokegama wetland, surface water pH was comparable in fall and spring.
Fall surface water samples contained the highest concentrations of SRP and all forms of N at the Pokegama wetland (Table 2, Fig. 7). Surface water NH4–N concentrations were particularly high in the fall, at least five times higher than NH4–N concentrations in either wetland during spring or summer (Fig. 7a and 7g).
There was a large spring pulse of TSS and VSS at the Pokegama wetland, presumably due to suspended clays carried in the turbid Pokegama River water (Table 4). The ratio of VSS to TSS was only 200 g kg-1 in the Pokegama riverbed in the spring, but was 900 g kg-1 in backwater organic zones in both wetlands in the summer.
Comparisons between Wetlands
There were few differences between the two wetlands in surface water characteristics (Table 5). In the spring, surface water concentrations of NO3–N were much higher at the Fond du Lac wetland than in the Pokegama wetland, whereas surface water concentrations of TP and TSS were lower. In the summer, surface water concentrations converged on similar values in the backwaters of both wetlands. Even the average summer NO3–N concentration, which was 25 times higher in the Fond du Lac riverbed than in the Pokegama riverbed, was similar in the two backwater areas. This convergence of values implies that common seasddonal mechanisms are controlling nutrient concentrations in surface waters overlying backwater areas. Summer concentrations of SRP and TN were higher in the Fond du Lac wetland than in the Pokegama wetland. The two wetlands were comparable with regard to pH, NH4–N, and VSS in surface water, regardless of season.
Effect of Within-Wetland Geomorphology
During the summer, water samples from riverbed sample points were quite different than those from backwater sample points, with higher pH and higher concentrations of NH4–N, SRP, TP, and TSS (Table 5). Summer NO3–N concentrations were nearly two orders of magnitude higher in the Fond du Lac riverbed (100 µg L-1) than in its backwater (2 µg L-1), but summer NO3–N concentrations were uniformly low (4 µg L-1) throughout the Pokegama wetland (Fig. 7b and 7h). Summer TN was also significantly higher in the Fond du Lac riverbed (1200 µg L-1) than in its backwater (1000 µg L-1: Fig. 7d).
Geomorphology influenced surface water less in the spring. Concentrations were significantly higher in the riverbed than in the backwater for NO3–N at Fond du Lac (Fig. 7b), and for TN and TP at Pokegama (Fig. 7j and 7k). Spring TSS concentrations were also higher in waters overlying riverbed and mineral backwater zones at Pokegama than in waters overlying organic backwater zones (Table 5, Tukey comparison).
Seasonal trends were often different in riverbed surface waters than in backwater surface waters (i.e., significant interactions between season and geomorphology, Table 2). In both wetlands, NH4–N concentrations increased between spring and summer in riverbed surface waters, but decreased in backwater surface waters (Fig. 7a and 7g). At the Fond du Lac wetland, NO3–N concentrations also increased between spring and summer in riverbed wetlands, but decreased in backwater wetlands (Fig. 7b). Seasonal increases in surface water SRP were greatest in riverbed zones at both wetlands, although increases occurred in backwater mineral areas as well (Fig. 7c and 7i).
Material Fluxes
Denitrification was measured on unamended samples and on samples amended by the addition of NO-3. There was virtually no denitrification in unamended samples, but sample amendment greatly stimulated denitrification, indicating that the process was NO-3 limited (Tables 4 and 6). Denitrification rates from amended soils collected in the summer were significantly higher than from soils collected in the spring (P < 0.0005), but denitrification from unamended soils was consistently low and did not vary seasonally (P = 0.235). Amended denitrification rates were an order of magnitude higher at Fond du Lac than at Pokegama in the spring and were also significantly higher in the summer (Table 4).
The effect of geomorphology on denitrification varied with treatment and season (Tables 4 and 6). There was no effect of geomorphology on unamended denitrification rates. In spring, amended denitrification was higher in riverbeds and levees than in backwater zones. In summer, amended denitrification was highest in levee and organic backwater zones. Given that organic backwater zones had the highest C content (Fig. 3a), this may infer that denitrification was also C limited in the summer.
Sedimentation rates were measured throughout the growing season in 1993 and 1994 (Fig. 8). Sedimentation rates were higher in riverbed than in backwater zones in both 1993 (P = 0.042) and 1994 (P < 0.0005), although there was considerable variability in measured rates (Fig. 8). Sedimentation in 1994 was significantly higher at Fond du Lac than at Pokegama (P < 0.0005). There was no significant difference in sedimentation rates between years (P = 0.332).
Hypotheses Tested
1. . Sedimentation Rates and Nutrient Concentrations in Overlying Floodwater Will Decrease with Distance from the River Channel
This hypothesis was true with regard to average sedimentation rates (Fig. 8). In those instances where there was a significant effect of geomorphology on the surface water chemistry of overlying floodwater, it was true that the highest average concentration consistently occurred in one or both riverbed zones (Table 5). However, significant differences between river water and backwater concentrations at one wetland often did not occur simultaneously at the other.
2. . Total Phosphorus Concentrations per Unit Volume of Soil Will Be Highest in Areas with High Sedimentation Rates
True. Total P concentrations and sedimentation rates were both highest in riverbed and levee zones (Fig. 3c and 8), and there was a low but significant correlation (r = 0.32) between soil TP and 1994 sedimentation rates when data from the two sites were combined.
3. . Soil Ammonium Concentrations Will Be Highest in Areas with High Organic Matter Content (Stimulating Mineralization) and Poor Aeration (Inhibiting Nitrification)
True. Ammonium concentrations were highest in sediments with high organic matter content and lowest in the well-aerated Pokegama levee.
4. . Soil Nitrate Concentrations Will Be Highest in Well Aerated Areas That Are Closest to the River, Due to Nitrate Inputs from Floodwaters and Low Denitrification Rates
Nitrate concentrations were consistently highest in the levees, which are both well aerated and close to the river, but nitrification appeared to be the main source of N, rather than NO-3 inputs from floodwaters.
5. . Denitrification Will Be Primarily Limited by Nitrate Availability, and Thus Realized (Unamended) Rates Will Be Much Lower than Potential (Amended) Rates
True. Nitrate supply was clearly limiting denitrification in all zones during spring and summer, as indicated by large increases in denitrification rates when samples were amended with KNO3 (Tables 4 and 6). Realized denitrification rates were often below limits of detection.
6. . Given Hypotheses 4 and 5, Denitrification Will Be Greatest in Areas with High Concentrations of Nitrate in the Overlying Water Column and on Levees
Denitrification was consistently high in zones with the highest porewater NO-3 concentrations (i.e., levees). Denitrification rates did not appear to be related to NO-3 concentrations in the overlying water column.
|
DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
1.72 g N 100 g-1 (five studies summarized by Johnston, 1991). Also, our values represent aboveground nutrient concentrations for whole plants rather than leaves, which tend to be more N rich than other plant parts. Sediment sorbed to plant stems may also have diluted measured N concentrations. Because wetlands contain abundant plant biomass, it is often assumed that plant uptake contributes substantially to their ability to affect water quality. Although plant species did exhibit geomorphic zonation within the two wetlands studied (Sersland et al., 1995), herbaceous plant biomass did not (Table 1). Furthermore, observed geomorphic differences in plant nutrient concentrations (e.g., higher plant P concentrations in riverbed zones) seemed to be a reflection of nutrient availability (i.e., high P concentrations in riverbed soils and surface water), rather than a mechanism of nutrient retention.
Observed soil concentrations of NO3–N in the riverine wetlands were comparable with those in a streamside wetland studied by Johnston (1993), but NH4–N concentrations were higher: the streamside wetland averaged only 18 g m-3, as opposed to average concentrations exceeding 60 g m-3 in organic backwater zones of the riverine wetlands studied (Fig. 5c and 5h). Average inorganic P concentrations in the streamside wetland (234 g m-3) were lower than the soil AFP concentrations observed in the riverine wetlands (Fig. 5e and 5j). Like the riverine wetlands, the streamside wetland contained a range of soil types, with surface organic matter contents ranging from 3 to 37 g 100 g-1 (Johnston et al., 1984b, 1997). Porewater SRP concentrations in the riverine wetlands were comparable with those reported by Johnston et al. (1995) in a wet meadow Fluvaquent (17 µg L-1 in 1986, 31 µg L-1 in 1987), and porewater NH4–N concentrations exhibited similar extreme temporal variation in the Fluvaquent (1220 µg L-1 in 1986, 120 µg L-1 in 1987).
Nutrient concentrations measured in porewater were much more spatially variable than those in soil extractions, as indicated by large 95% confidence intervals around the geomorphic zone means (Fig. 6). Nitrate concentrations in levees were particularly variable (Fig. 6b and 6e), probably due to small pockets of aerobic and anaerobic conditions within levee soils that influence rates of nitrification and denitrification. This finding suggests that soil extractions may be a more integrative measure than porewater measurements. Nutrient concentrations in surface water tended to be the least spatially variable of the measurements made (i.e., small 95% confidence intervals around the geomorphic zone means).
Nutrient Dynamics by Geomorphic Zone
Riverbed
Riverbed zones are integrally connected with the rivers that supply them, such that waterborne materials can pass freely into and out of them. As a result of this free hydrologic exchange, inorganic nutrients and TSS are generally more concentrated in the surface waters of riverbed zones than they are in backwater zones. In our study, all water chemistry parameters except VSS attained their highest summer concentrations in one or both of the riverbed zones.
Given that rivers differ in water chemistry, the water chemistry of riverbed zones also varies from river to river. In our study, the water chemistry of riverbed zones on the St. Louis (Fond du Lac study site) and Pokegama Rivers was quite different, despite their proximity and common climate. The two riverbed zones fell at opposite extremes of high and low concentration for spring TP and TSS and summer NO3–N and TN. These extremes indicate the dynamic effects of river water inputs to riverbed wetlands, such that their water chemistry may be influenced more by upstream events than by in situ processes.
Although the flow of river water imports materials to riverbed wetlands, it clearly exports them as well. The export of fine sediment and organic matter from riverbed wetlands is inferred by their high sand content, low organic matter and C content, and low TN content (Fig. 2 and 3). Net retention occurs when high exports are offset by even higher inputs, as was the case for sedimentation. Although it seems paradoxical that riverbed wetlands were the locus of maximum sedimentation given that they are the geomorphic feature most susceptible to erosion, it makes sense given their high input rates from river water.
Riverbeds are zones of high P concentration, with some of the highest values for soil TP, vegetation TP, soil AFP, summer porewater SRP, and summer surface water SRP. Riverbeds receive large inputs of soluble P via river water and particulate-sorbed P via sedimentation. However, P sorption is lower in riverbed soils because of their high sand content, which was confirmed by separate sorption experiments (Bridgham et al., 2001). Thus, sedimentation, rather than sorption, is the primary P retention mechanism in riverbed zones.
Levee
Levees are slightly elevated above the water table, which promotes oxidation reactions, thereby depleting S- concentrations and increasing NO-3 concentrations. In our study, levees were hot spots of NO3–N, with porewater NO3–N concentrations orders of magnitude higher than elsewhere in the wetland. The NO3–N was generated by processes internal to levees, rather than imported from allochthonous sources, because river water concentrations of inorganic N were extremely low. We attribute these high NO3–N concentrations to the nitrification of NH4–N under the aerobic conditions that normally occur in levees. The high NO3–N concentrations in levees are probably not the result of a lack of denitrification, because denitrification potentials in the levees were comparable with or greater than denitrification potentials elsewhere in the wetlands. Factors that possibly could have limited in situ denitrification rates, however, are the lack of anaerobic microsites or C sources in the levee soils.
Small differences in elevation between the two levees had large effects on their inorganic N dynamics. Although the Pokegama levee was elevated above the water table throughout the year, the summer flooding of two-thirds of the Fond du Lac levee resulted in much different summer trends for NH4–N and NO3–N in the two wetlands. Thus, the position of the water table relative to the geomorphic structure of the levee, rather than the presence of the levee per se, had the greatest effect on inorganic N concentrations within the levee.
Like riverbeds, levees contain high concentrations of soil TP and summer porewater SRP. However, the two levees were different with regard to AFP. The Pokegama levee had the lowest AFP of any geomorphic feature in spring and summer, as well as the least soil moisture and the most alkaline soil pH, and had significantly less silt than the Fond du Lac levee. Using the data from all grid points, AFP was significantly (P < 0.05) negatively correlated with soil pH and positively correlated with silt percentage (Bridgham et al., 2001), which may explain the differences between levees.
In separate P sorption experiments (Bridgham et al., 2001), a Fond du Lac levee soil (the only levee soil tested) was found to be an outlier relative to riverbed and backwater soils. The levee soil released P when dilute P solutions (0.5–1.0 µmol L-1) were added; riverbed and backwater soils sorbed P at these concentrations. The levee soil also had a higher proportion of microbial P uptake when P solutions of 0.5 to 3.2 µmol L-1 were added to soil samples. Although the results from one sample point are by no means conclusive, they do imply that the P dynamics of levees may be unique relative to other geomorphic zones in riverine wetlands, in that microbial uptake processes may be more important than geochemical sorption processes in controlling soil P dynamics.
Levees form a physical barrier separating backwaters from the river mainstem. Unlike human-constructed levees, natural levees are discontinuous, reducing the mixing of river water and backwater, but not preventing it. Given that the Pokegama backwater received river water only through a narrow (20 m) levee opening, whereas the Fond du Lac backwater was bounded by a shallow bay of the St. Louis River, we expected that poorer exchange of surface water at the Pokegama site would increase differences in the chemistry of its river water vs. backwater, which would be indicated by a significant effect of site x geomorphology interaction. This was the case in spring, when TP and TSS were much higher in river water than in backwater only at the Pokegama site (see Tukey groupings in Table 5). However, in summer the differences in chemical concentrations between river water and backwater were as great or greater at the Fond du Lac site as they were at the Pokegama site. Thus, it appears that even discontinuous natural levees can effectively separate waters having different chemical characteristics.
Backwater
Backwaters in the two wetlands had surprisingly similar summer surface water chemistry, despite often substantial differences in substrate and river water chemistry. For example, summer backwater pH and NO3–N concentrations were comparable at both wetlands, despite pronounced differences in river water pH and NO3–N concentrations. The two backwater areas also had comparable summer surface water NH4–N, TP, TSS, and VSS. This seasonal convergence of chemistry implies that biotic processes common to the two backwater areas override their substrate differences.
Of the four geomorphic zones, organic and mineral backwater areas were most similar to each other. They differed by definition in C and organic matter content, and also differed in soil properties related to organic matter (i.e., soil moisture, bulk density). Soils of organic backwater zones contained less TP than other soils, but contained AFP and porewater SRP concentrations comparable with those in mineral backwater areas. Organic backwater zones were also similar to mineral backwater zones in water depth, soil TN, and soil and porewater NO3–N concentrations.
Despite these similarities, organic backwater zones were different than other zones in spring NH4–N concentrations (soil and porewater), in summer surface water pH, and in summer denitrification potential. The high organic matter and C content of these soils probably increased N mineralization, released organic acids (Eshleman and Hemond, 1985), and contributed C for NO3–N reduction to cause these effects. Because all soil nutrient concentrations were expressed on a per-volume basis, there were fewer differences between organic backwater and other soils than would have occurred if expressed on a per-mass basis.
Backwaters in both wetlands exhibited a pronounced seasonal decrease in soil S- and concurrent increase in soil NO3–N concentrations. Most other nutrient concentrations remained seasonally stable in backwater soils, particularly in organic backwater zones. Soil pH and soil NH4–N changed little between spring and summer, and changes in soil AFP and porewater SRP concentrations were small relative to those in riverbed and levee zones.
Inter-Wetland Differences
The Pokegama and Fond du Lac wetlands were intentionally chosen to represent different mineral substrates. Given this a priori selection, it was surprising that there were so few inter-wetland differences. Those inter-wetland differences that did occur were mostly the result of differences in riverbeds, which received different inputs from their respective rivers, and in levees, which responded differently due to differences in elevation. These findings imply that within-wetland variation is more important to nutrient cycling dynamics than is inter-wetland variation. The convergence of properties in backwater areas of the two wetlands also implies that differences in mineral soil texture have less bearing on nutrient cycling processes there, particularly as the long-term accumulation of organic matter reduces initial substrate differences.
|
CONCLUSIONS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
1. Intra-wetland variation is greater than inter-wetland variation.
2. Experimental designs of future studies must incorporate this variability to legitimately extrapolate to the entire wetland.
3. Given the small differences between the organic and mineral backwater zones, hydrologic zones (i.e., riverbed, levee, backwater) are the major source of intra-wetland variability.
4. This implies that hydrologic factors are more important than physical soil characteristics in determining nutrient cycling, so studies of nutrient cycling in riverine wetlands should focus on an experimental design that subdivides the wetland into appropriate hydrologic zones.
5. More research needs to be done on the interplay between hydrology and nutrient cycling in riverine wetlands.
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Received for publication August 13, 1999.
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
