a Environmental Science and Public Policy Program, MS 5F2, George Mason Univ., Fairfax, VA 22030-4444
b School of Forestry, Auburn Univ., Auburn Univ., AL 36849
c Environmental Science and Public Policy Program, MS 5F2, George Mason University, Fairfax, VA 22030-4444
* Corresponding author (email@example.com )
Studies of how flooding affects P availability in natural floodplains are rare. We examined the effects of artificial flooding on P availability in a Georgia floodplain forest. We hypothesized that P availability would increase with flooding, because of the flooding-induced solubilization of phosphate minerals. Field mesocosms (n = 4 per treatment) were flooded with river water according to one of four treatments over 6 mo: (i) continuously flooded; (ii) flooded for 3 mo and then drained; (iii) flooded for 2 mo, drained for 1 mo, and repeated; and (iv) nonflooded control. Two additional sets of 3-mo flooded–drained mesocosms (n = 4 per set) received added P or N (1 and 10 mg L-1, respectively) with flooding. Soils were collected monthly from both inside and outside of the mesocosms and analyzed by Hedley fractionation; anion-exchange resins (AER) were used to estimate P availability in situ. As indexed by daily supply to AER, P availability was significantly greater in flooded versus control soils, and decreased significantly following drainage, in all treatments at some time during the study. Total P supply to AER was significantly greater in flooded versus control mesocosms regardless of treatment. No significant changes were observed in Fe/Al phosphate fractions. Microbial P was significantly lower in flooded versus control mesocosms during the first 3 mo of flooding and decreased significantly over time in two treatments. In this natural floodplain, biological processes are a more probable explanation for flooding-induced increases in P availability than solubilization of mineral phosphates.
Abbreviations: AER, anion-exchange resins • CF, continuously flooded treatment • dwe, dry weight equivalent • FD, flooded–drained treatment • LOI, loss on ignition • kp, microbial biomass P recovery rate • MC, mesocosm control • NaOH I, the first NaOH extraction in the Hedley fractionation scheme • NaOH II, a second NaOH extraction after ultrasonification of soil particles • NC, nonmesocosm control • +N, flooded–drained treatment with N additions • OM, organic matter • PF, periodically flooded treatment • +P, flooded–drained treatment with P additions • Pi, inorganic P • Po, organic P • Pt, total P
Soil Science Society of America Journal 65:1293-1302 (2001).
FLOODPLAIN FORESTS have been shown to play an important role in improving downstream water quality (National Research Council, 1995; Brinson and Rheinhardt, 1996). This capability results in part from a landscape position that can favor the flow of both upstream (via overbank flooding) and upslope waters (via riparian transport) through the floodplain at various times during the year (Brinson, 1993; Lockaby and Walbridge, 1998). Interest in the ability of floodplain forests to remove suspended and dissolved constituents from the water column has generated extensive research (Brinson, 1977; Mitsch et al., 1979; Yarbro, 1983; Brinson et al., 1984; Elder, 1985; Mendelssohn and Burdick, 1988) and several reviews (Lowrance et al., 1984; Faulkner and Richardson, 1989; Kuenzler, 1989; Johnston, 1991; Walbridge and Lockaby, 1994; Lockaby and Walbridge, 1998). Phosphorus retention is of particular concern because P often limits primary production in downstream aquatic systems. Excess inputs of dissolved P to these systems can cause stream eutrophication and fish kills (NRC, 1995). Despite this concern, detailed studies of the effects of flooding on the distribution and availability of P in floodplain soils are rare and limited to alkaline soils (Fabre et al., 1996).
When upland soils are flooded for rice cultivation, P availability has been shown to increase because of the release of P from geochemical pools: (i) the reduction and dissolution of Fe (III) phosphates; (ii) the hydrolysis and dissolution of Fe and Al phosphates; or (iii) the release of clay-associated phosphates through anionic exchange (Ponnamperuma, 1972; Gambrell and Patrick, 1978). In contrast, more recent research has shown that in some systems, flooding can cause P availability to decrease by favoring the formation and persistence of amorphous (poorly crystalline) Fe and Al minerals (Kuo and Mikkelsen, 1979; Sah and Mikkelsen, 1986; Sah et al., 1989). Because their adsorptive surface area per unit soil volume is larger than more crystalline forms (Parfitt, 1989; Schwertmann and Taylor, 1989), these amorphous minerals tend to dominate soil P sorption reactions when present in significant amounts.
Several other factors might cause P availability to increase with flooding. In agricultural or urban areas, floodwaters themselves can carry significant loadings of available P, resulting from point and nonpoint source pollution within the watershed (Carpenter et al., 1998). Phosphorus availability might also increase following flooding because of a decrease in biological (plant and microbial) P demand under anaerobic versus aerobic conditions (Gambrell and Patrick, 1978; Schlesinger, 1997; Mitsch and Gosselink, 2000). Labile P can also be released from microbial biomass following flooding because of both the lysis of aerobic microorganisms under anaerobic conditions and the activities of facultative anaerobes specially adapted to fluctuating oxic/anoxic environments (Boström et al., 1985; Fleischer, 1985; Davelaar, 1993; Khoshmanesh et al., 1999).
The objective of this study was to examine the effects of flooding on P availability in a forested floodplain ecosystem along the Ogeechee River, Georgia. We hypothesized that P availability would increase with increased flooding duration and periodicity, primarily because of the flooding-induced release of P from Fe and Al phosphates, as observed in upland soils flooded for rice cultivation. Flooding duration and periodicity were manipulated experimentally in field mesocosms artificially flooded with river water. Previous research at this site emphasized organic matter (OM) and bacterial metabolism in the river and export of OM and bacteria from the floodplain (Findlay et al., 1986; Meyer, 1986; Cuffney and Wallace, 1987; Edwards and Meyer, 1987; Cuffney, 1988; Meyer and Edwards, 1990; Carlough and Meyer, 1991; Pulliam, 1992; Wainright et al., 1992). More recent research has focused on the effects of flooding on biogeochemical processes (Lockaby et al., 1996a, 1996b; Darke et al., 1997; Darke and Walbridge, 2000).
The study site was a floodplain forest on the east bank of the Ogeechee River, Georgia (32°08'N, 81°23'W) (Fig. 1). At this point, the Ogeechee is a low-gradient (19.6 cm km-1) sixth-order alluvial blackwater river (Wharton, 1970). The river gets its distinctive color from dissolved organic C, leached from sandy soils typical of southeastern Coastal Plain uplands (Meyer, 1986). The Ogeechee River runs 390 km through the Georgia Coastal Plain to the Atlantic Ocean (Cuffney and Wallace, 1987) and is considered relatively unpolluted (Meyer and Edwards, 1990). Annual flooding occurs during the winter and spring months, primarily from January to April (Benke and Meyer, 1988). Water depths in the floodplain can vary throughout the year from 0 to >2 m (Cuffney and Wallace, 1987). The pH of the river water ranges from 6.6 to 8.1 annually (Cuffney and Wallace, 1987). Average water temperature is 19°C; summer temperatures (June to August) average 28°C (Findlay et al., 1986). Average total P (Pt) concentration in the Ogeechee River is 59 µg L-1, with an average N/P ratio of 10 (Meyer, 1986; Benke and Meyer, 1988).
Soils are loams of the Chastain series (fine, mixed, semiactive, acid, thermic Fluvaquentic Endoaquepts) and silty clay loams of the Tawcaw series (fine, kaolinitic, thermic Fluvaquentic Dystrudepts) (M. Thomas, USDA Soil Scientist, Effingham Co., GA, personal communication). As estimated by loss on ignition (LOI), soil OM content ranges from -1 on levees to as much as 500 g kg-1 in the soils of stagnant back swamps adjacent to the upland (Cuffney, 1988). Higher elevation ridges are dominated by Quercus laurifolia M., Carpinus caroliniana W., Pinus spp., and Arundinaria gigantea W. Muhl.; lower elevation swales are dominated by Quercus nigra L., Taxodium distichum L. Richard, Nyssa sylvatica M., Nyssa aquatica L., Liquidambar styraciflua L., and Sabal minor (Jacquin) Persoon (Radford et al., 1968; Lockaby et al., 1996a).
The effects of flooding on soil P pool sizes and P availability were addressed in a 6-mo in situ mesocosm study. Mesocosms (n = 24) were constructed from open-ended plastic cylinders (0.45-m diam. and 1.0-m height) inserted into the floodplain substrate in March 1992 to a depth of 0.45 m, in a 30 by 60 m area located near the river (for a more detailed description, see Lockaby et al., 1996a). Mesocosms were arranged in a randomized complete block design with blocking based on surface soil characteristics (i.e., percent OM, extractable P, and pH) (Lockaby et al., 1996a). River water was pumped to the mesocosms at a constant rate according to the following four treatments (n = 4 mesocosms per treatment): (i) continuously floooded for 6 mo (CF); (ii) flooded–drained (FD): flooded for 3 mo and then drained for 3 mo; (iii) periodically flooded (PF): flooded for 2 mo, drained for 1 mo, flooded for 2 mo, and drained for the final month; and (iv) nonflooded control (mesocosm control; MC). Two more sets of mesocosms (n = 4 mesocosms per set) were flooded according to the FD regime, with one set receiving elevated P (+P) and one set receiving elevated N (+N). Constant drip rates with target concentrations of 1 and 10 mg L-1 for PO4–P and NH4–N, respectively, were maintained with peristaltic pumps (for a more detailed description, see Lockaby et al., 1996a). An elevational difference of 10 cm between intake and outflow spouts ensured continuous water movement through each mesocosm, maintaining dissolved O2 levels in overlying floodwaters similar to those of river waters (5–6 mg L-1).
Soil Collection and Analysis
In January 1992, soil samples (n = 20) were collected to a depth of 15 cm from the area where mesocosms would later be established to gain baseline information on soil P pool sizes and P availability. In June 1992 (at the beginning of the study period), the late winter-early spring flooding season had ended and no standing water was present in the mesocosm study area. Immediately following treatment initiation in June 1992, and at monthly intervals thereafter for 6 mo, single soil samples were collected from each mesocosm to a depth of 10 cm. During this period, soil samples (n = 4) were also collected, at random, from locations outside the mesocosms to examine the effects of mesocosm construction on soil P pool sizes and P availability (nonmesocosm control [NC]). All soils were collected using a 5.2-cm-diam. (Al) bulb planter. Soil sampling removed from each mesocosm over the 6-mo study period. Collected soil samples were deposited in polyethylene bags, placed in coolers filled with ice, and stored at 4°C upon returning to the laboratory. Soils were composited, removing coarse leaf and root material, and 10- to 20-g portions were dried to a constant mass at 80°C to estimate gravimetric moisture content. Soils collected in January 1992 were analyzed for pH in a 1:2 slurry of soil/deionized water, soil texture (Bouyoucos, 1962), and percent OM by LOI at 550°C overnight in a muffle furnace (Lim and Jackson, 1982).
All soils were analyzed for P fractions using a modification of the Hedley et al. (1982a)(1982b) fractionation procedure. In this procedure, soil P fractions are sequentially extracted on the basis of their relative solubilities in water, alkali, and acid. Readily extractable forms (e.g., water-soluble, resin-extractable, and NaHCO3-extractable P) are considered immediately available to plants (Agbenin and Tiessen, 1995; Cross and Schlesinger, 1995); less readily extractable forms (e.g., NaOH-extractable P) are considered available to plants over a growing season (Agbenin and Tiessen, 1995); and recalcitrant forms (e.g., NaOH II–extractable P [a second NaOH extraction after ultrasonification of soil particles], hydrochloric acid [HCl]-extractable P, and residual P) are considered available to plants only over long periods of time, if ever (Cross and Schlesinger, 1995).
Duplicate (n = 2) 0.5-g dry weight equivalent (dwe) field-moist subsamples of each soil sample were placed in 50-mL screw-cap centrifuge tubes. Extractions for labile P (resin-extractable, water-soluble, NaHCO3-extractable, and chloroform [CHCl3]-fumigated/NaHCO3-extractable P) were initiated within 30 h of soil collection. In this modified Hedley procedure, the fractionation sequence was initiated with the estimation of microbial biomass P by CHCl3 fumigation (CHCl3–fumigated/NaHCO3-extractable P) to minimize potential negative effects of resin extraction on microbial biomass P. Resin-extractable and NaHCO3-extractable P were determined on separate 0.5-g dwe subsamples (n = 2) of each soil sample, that is, subsamples that did not proceed through the remainder of the Hedley fractionation scheme.
In preparation for soil extraction, AER bags (0.4-g air dried Dowex 1 by 8, >0.450-mm-diam. Cl-form AER beads encased in mesh nylon stocking bags) were converted to 75% HCO3 form by incubating each bag in 30 mL of 0.5 M NaHCO3 for two half-hour sessions, using a fresh solution each time. Soil samples were shaken with resin bags in 30 mL of deionized water for 16 h using a reciprocating water bath shaker. Following shaking, resin bags were removed and supernatants were centrifuged for 20 min at 1200 x g (3000 rpm). The supernatant solution was vacuum filtered through 45-µm membrane filters (Supor 450, Pall Life Sciences, Pensacola, FL) and digested with ammonium peroxydisulfate in an autoclave for 30 min at 120°C and 100 kPa (15 lb in-2) of pressure (Grasshoff et al., 1983) to estimate water-soluble organic P (Po). Resin-extractable inorganic P (Pi) was estimated by incubating resin bags for 16 h in 30 mL of 0.5 M HCl per bag and then shaking for an additional 0.5 h. Incubation is required to release CO2 to prevent subsequent interference with bubble separation of samples during autoanalysis. Following extraction, resin bags were converted back to 75% HCO3 form for reuse.
Sodium bicarbonate–extractable Pi was estimated by shaking 0.5-g soil subsamples with 30 mL of 0.5 M NaHCO3 for 16 h, then centrifuging and filtering as described above. To estimate P released and subsequently readsorbed by soil particles during NaHCO3 extraction, separate 0.5-g subsamples (n = 2) were also shaken for 16 h with 0.15 mL KH2PO4 (250 µg P mL-1) and 30 mL of 0.5 M NaHCO3, then centrifuged and filtered as above (Brookes et al., 1982). Both NaHCO3-extractable P and CHCl3-fumigated/NaHCO3-extractable P values were adjusted for P readsorption during extraction following Brookes et al. (1982).
To initiate the Hedley fractionation scheme, soils were treated with 0.5 mL of liquid CHCl3, and centrifuge tubes were then capped, incubated for 16 h under a fumigation hood, uncapped, and then incubated for an additional 2 h. Soils were then shaken with 30 mL of 0.5 M NaHCO3 for 16 h, and centrifuged and filtered as described above. Microbial biomass P was estimated as the difference in 0.5 M NaHCO3-extractable Pi in CHCl3-fumigated versus nonfumigated soils, using a recovery rate (kp) of 0.4, which has been found to represent the average recovery of microbial P by this technique for a wide range of soils (Brookes et al., 1982; Hedley and Stewart, 1982; Walbridge, 1991).
Following CHCl3–fumigation and 0.5 M NaHCO3 extraction, soil residues were sequentially extracted with (i) 30 mL of 0.1 M NaOH (NaOH I P); (ii) 30 mL 0.1 M NaOH following 2 min of ultrasonification (with centrifuge tubes in ice) (NaOH II P); and (iii) 30 mL of 1.0 M HCl (HCl P). Following the 16-h extractions, all extracts were centrifuged and filtered as described above. Sodium hydroxide I and II extracts were centrifuged at 4°C for 90 min to facilitate filtration. All extracts were adjusted to pH 2.5 and analyzed for Pi by the method of Murphy and Riley (1962), using a Technicon Autoanalyzer II (Technicon Instrument Corporation, Tarreytown, NY) (method number 696-82W; Bran and Luebbe, 1989). Colored extracts were subsequently reanalyzed for potential color interference after removal of ascorbic acid from the autoanalysis reagent stream (i.e., preventing formation of the molybdate blue color in the presence of Pi). Samples registering above the newly established (lower) baseline were noted and differences were attributed to color interference and were factored into concentration calculations. The Po concentration of 0.5 M NaHCO3 extracts (including CHCl3-fumigated/NaHCO3-extractable P) and 0.1 M NaOH extracts was estimated as the difference between Pt following autoclave digestion of supernatants, using methods described for determinating water-soluble Po and supernatant Pi concentrations. Microbial Po was estimated as the difference between CHCl3-fumigated/NaHCO3-extractable Po and NaHCO3-extractable Po. Hydrochloric acid supernatants were not analyzed for their Po content because previous investigators have found Po concentrations in these extracts to be negligible (Hedley et al., 1982a). Following HCl-extraction, residual soils from the fractionation sequence were digested in a Technicon block digester (Technicon Instrument Corp., Tarreytown, NY) with concentrated sulfuric acid (H2SO4) and 30% hydrogen peroxide (H2O2) for at least 30 min at 360°C (Haynes, 1980; Lowther, 1980). The resulting supernatant was analyzed for Pt content (residual P) using a Technicon Autoanalyzer II (method number 696-82W; Bran and Luebbe, 1989). Nonextracted soils (duplicate 0.200-g subsamples) were digested similarly to estimate Pt, for comparison with the sum of the Hedley fractions. Digests of soils collected in January 1992 were also analyzed for total N content (method number 696-82W; Bran and Luebbe, 1989).
In Situ Anion-Exchange Resins
Anion-exchange resin bags (8.0 g of air dried Dowex 1 by 8, >0.450-mm-diam. Cl-form AER beads encased in nylon stocking material and converted to 75% HCO-3 form using methods described above) were buried and replaced in each mesocosm (n = 1 per mesocosm) at monthly intervals as an in situ estimate of P availability. Resin bags (n = 4) were also buried and replaced at monthly intervals in soils outside the mesocosms. Resins exchange HCO-3 for PO3-4 in the soil solution in a manner similar to plant roots; P availability indexed by in situ resin bags has been shown to correlate positively with P availability to plants (Walbridge, 1991; Giblin et al., 1994). Phosphate was extracted from resin bags using methods described above for 0.4-g resin bags, using 100 mL of extracting solution per resin bag. Daily P supply to in situ AER (µg d-1) was calculated as the amount of P absorbed by each resin bag divided by the number of days it was buried in the soil. Total P supply to in situ AER is the amount of P absorbed by each resin bag during the time it was buried in the soil.
The method of restricted maximum likelihood (PROC MIXED) was used to examine differences in soil P pool sizes and P availability as a function of time (SAS, 1996). Autocorrelation was taken into account using a repeated measures statement within the procedure. Individual significant differences between sampling periods within each treatment were identified using the method of least squares means with a Tukey adjustment (SAS, 1996). Differences in soil P pool sizes and P availability as a function of treatment were examined using ANOVA (SAS, 1996). Tukey's studentized range simultaneous comparison procedure was used to identify significant differences between treatments within each sampling period (SAS, 1996). In situ AER P data were log-transformed before statistical analysis to meet the assumption of normality required by the ANOVA method. Differences in total P supply to in situ AER in treatment versus control soils were examined using ANOVA for unbalanced sample sizes (PROC GLM) (SAS, 1996). The method of ANOVA with contrast statements was used to compare microbial biomass P in common flooded treatments versus controls for each month of flooding (SAS, 1996).
General Soil Characteristics
Soils in the mesocosm field were slightly acid (pH = 4.9) sandy loams with an OM content of 91 g kg-1, and a total N/P ratio of 7.3 (Table 1). On the basis of their contribution to total Hedley-extractable P (i.e., the sum of NaHCO3 Pi and Po, microbial P, NaOH I and II Pi and Po, HCl P, and residual P fractions), dominant soil P fractions were microbial P (29%), NaOH I Po (17%), and NaOH I Pi (14%) (Table 1). Total P estimated by the Hedley procedure (375.5 mg kg-1) was comparable to that estimated by persulfate/peroxide digests (349.7 mg kg-1) (Table 1).
Soil Phosphorus Pool Sizes and Phosphorus Availability
As indexed by daily supply to in situ AER, P availability was significantly greater in flooded versus control soils, and decreased significantly following drainage, in each of the five flooding treatments at some time during the 6-mo study (Fig. 2A–G). Total P supply to in situ AER was significantly greater in flooded versus control mesocosms regardless of treatment (Table 2).
Resin-extractable Pi increased significantly as a function of both nutrient addition and flooding in flooded-drained soils (Fig. 3E–G), and decreased significantly following drainage in flooded–drained soils receiving nutrient additions (Fig. 3F and G). Despite significant changes in resin-extractable Pi over time in flooded–drained soils, resin-extractable Pi was only occasionally significantly different in flooded versus control soils (Fig. 3). Remaining Hedley soil P fractions (water-soluble, NaHCO3, NaOH I and II, HCl, and residual) did not differ significantly with either treatment or time (Wright, 1998). Thus we report only average values for these parameters (Table 1).
Microbial biomass P decreased significantly over time in two flooding treatments: FD (157.2 ± 39.8 kg P ha-1 in June versus 57.9 ± 13.2 kg P ha-1 in August) and +N (180.3 ± 7.4 kg P ha-1 in June versus 39.7 ± 6.1 kg P ha-1 in September) (Fig. 4E and G). Similar trends were observed in other flooding treatments, although they were not statistically significant (Fig. 4A, B, F). However, microbial biomass P was significantly lower in flooded versus control mesocosms during the first three months of flooding (Table 3). Microbial biomass Po changed little over the course of the experiment (Wright, 1998).
Total P in Ogeechee surface soils (349.7 ± 21.3 and 375.5 ± 29.0 mg kg-1; Table 1) was at the low end of Pt values reported for wetlands (300–1400 mg kg-1) by Faulkner and Richardson (1989); soil N/P ratios were within the range reported for forested wetlands in the southern USA by Bedford et al. (1999). Resin-extractable Pi in Ogeechee floodplain soils (9.5 ± 1.4 mg kg-1; Table 1) was similar to values cited by Stanturf and Schoenholtz (1998) for extractable P in blackwater river floodplain soils (10–11 mg kg-1; extraction method not defined by authors). Microbial P (CHCl3-fumigated/NaHCO3-extractable P less NaHCO3-extractable P and adjusted with a kp of 0.4) concentrations may seem high, both in absolute amounts and in relation to other fractions (Table 1). However as a percentage of total P, microbial P (29.2%) was comparable to values reported for NC Coastal Plain bay forest soils (37.4%) (Walbridge, 1991) and Appalachian forest soils (18–24%) (Walbridge et al., 1991), and was markedly less than microbial P in wetland wastewater treatment systems (60%) (Lee et al., 1975; Sloey et al., 1978). Microbial P in Ogeechee floodplain soils (109.6 ± 10.7 mg kg-1; Table 1) was above the range reported by Brookes et al. (1982) for nonnative (added) microorganisms in a variety of soils, including those under native grasses, deciduous forest, and cultivated crops (5.3–72.3 mg kg-1), but comparable to levels of microbial biomass P reported by Joergensen and Scheu (1999) for a mineral forest soil (86–114 mg kg-1).
We hypothesized that P availability would increase with increased flooding duration and periodicity, because of the flooding-induced release of P from Fe and Al phosphates. Phosphorus availability did increase in response to artificial flooding. As indexed by daily supply to in situ AER, P availability was either significantly greater in flooded versus control soils, and/or decreased significantly following drainage, in each of the five flooding treatments at some time during the 6-mo study (Fig. 2), while total P supply to in situ AER was significantly greater in flooded versus control mesocosms regardless of treatment (Table 2). However, during the course of the experiment, we observed no significant changes in Fe and Al (NaOH I and II) P in any treatment (Wright, 1998) that suggested the observed increases in P availability were the result of the solubilization of P from phosphate minerals. Similarly, Darke et al. (1997) found no significant changes in soil Fe or Al chemistry in response to flooding in this same experiment.
Previous research on upland soils flooded for rice cultivation has suggested that increases in P availability following flooding can be attributed to the reduction and dissolution of Fe (III) phosphates, the hydrolysis and dissolution of Fe and Al phosphates, or the release of clay-associated phosphates through anionic exchange (Ponnamperuma, 1972; Gambrell and Patrick, 1978). Because we observed no flooding-induced changes in Hedley P fractions representing Fe and Al phosphates in these Ogeechee floodplain soils, there is little evidence to suggest that observed increases in P availability with flooding (Fig. 2) were caused by the release of P from geochemical sources.
Alternative explanations for the observed increases in P availability include: (i) P inputs received in river water; (ii) a decrease in biological P demand under anaerobic conditions; or (iii) release of labile P from microbial biomass. With average Pt concentrations in Ogeechee River water of 59 µg L-1 (Meyer, 1986) and the low flow rates used in this artificial flooding experiment, river water is an unlikely explanation for the increases in P availability observed with flooding. Similarly, because the quality of river waters used to flood our mesocosms was uniform across treatments, flooding-induced increases in P availability associated with river water should have been more uniform across treatments than was observed (Fig. 2).
The decrease in biological P demand that occurs as aerobic microorganisms and plants become dormant under anaerobic conditions (Gambrell and Patrick, 1978; Schlesinger, 1997), at least partially explains the increases in P availability observed with flooding. While investigating the effects of flooding on decomposition dynamics in another aspect of this study, Lockaby et al. (1996a) found that significant P release from litter only occurred after drainage, suggesting that microbial activity was inhibited by flooding. Microbial N demand is known to decline under anaerobic conditions (Gambrell and Patrick, 1978). Since microbial N/P ratios are relatively constant over a wide range of soil conditions (Schlesinger, 1997), changes in microbial P demand should be consistent with those observed for N. In addition, many plants are known to tolerate short-term flooding through dormancy (cf., Mitsch and Gosselink, 2000).
A third explanation might be the release of labile P from microbial biomass following flooding, because of either the lysis of aerobic microorganisms under anaerobic conditions or the activities of facultative anaerobes specially adapted to fluctuating oxic and anoxic environments (Fleischer, 1985; Davelaar, 1993). In both wastewater treatment plants (Wentzel et al., 1986) and lake sediments (Boström et al., 1988; Gächter et al., 1988; del Carmen Doria-Serrano et al., 1992; Davelaar, 1993; Gächter and Meyer, 1993; van Veen et al., 1993; Waara et al., 1993), facultative anaerobes have been shown to accumulate Pi in polymer form ("poly-P") under aerobic conditions, hydrolyzing this stored poly-P under anoxic conditions when oxidative phosphorylation is no longer possible. The hydrolyzed Pi builds up within the cells and is released to the environment via diffusion (Wentzel et al., 1986). While poly-P accumulation has been of particular interest to the wastewater treatment industry because of pollutant P removal possibilities, microbes capable of these activities are also common in natural systems, including surface waters and soils (Gächter and Meyer, 1993). Boström et al. (1985) found a large decline in lake sediment residual P (defined as organic and inert P) with anoxia, but no significant fluctuations in NaOH–extractable P, and attributed the decline to a sediment bacteria P release. A recent laboratory study (Khoshmanesh et al., 1999) of wetland sediments also suggested a microbial accumulation of P in aerobic conditions and a subsequent release of P in anaerobic conditions when a proper fermentation product, such as acetate, was present.
In our Ogeechee soils, we observed significant decreases in microbial biomass P over time in two flooding treatments (Fig. 4E and G), and significantly lower microbial P concentrations in flooded versus control mesocosm soils during the first 3 mo of flooding, regardless of treatment (Table 3). Because experimental flooding treatments were initiated in late June, aerobic and facultative anaerobic soil microbial communities would probably have been well-developed.
In these natural floodplain soils, P availability increased in response to flooding. We found no evidence to suggest that flooding-induced increases in P availability were caused by the release of P from Fe and Al phosphates. We did observe significant decreases in microbial biomass P over time in two treatments, and significantly lower microbial P concentrations in flooded soils during the first three months of flooding, regardless of treatment. In natural wetland soils, where periodic flooding is a major factor driving soil pedogenesis, Fe and Al phosphates are probably liberated from parent materials by flooding fairly early during soil development. In this natural floodplain wetland, biological mechanisms provide a more probable explanation for observed flooding-induced increases in P availability.
The authors thank T.M. Williams for providing baseline water table elevations, T. Carpenter for conducting soil physical analyses, M. Thomas for providing unpublished USDA Soil Survey information for Effingham County, Georgia, and C.D. Sutton for providing assistance with statistical analyses. S.L. Beach, R.C. Jones, J. Lawrey, and two anonymous reviewers provided helpful comments on an earlier draft of this manuscript. This research was supported in part by a USDA Competitive Research Grant to Mark R. Walbridge.
Received for publication August 23, 1999.
Fig. 1. Map of the study site on the east bank of the Ogeechee River, Georgia (from Darke and Walbridge, 2000, with permission).
Fig. 2. The effects of artificial flooding treatments on soil P availability as measured by daily P supply rate to in situ AER bags in field mesocosms. Brackets indicate periods of flooding. (A) CF = continuously flooded; (B) PF = periodically flooded; (C) MC = mesocosm control; (D) NC = nonmesocosm control; (E) FD = flooded–drained; and (F) +P and (G) +N, flooded–drained treatments with nutrients added. Lowercase letters indicate significant differences over time within a treatment (abc) (based on the method of least squares means with a Tukey adjustment [family error rate of P
Fig. 3. The effects of artificial flooding treatments on resin-extractable Pi in field mesocosms. Brackets indicate periods of flooding. (A) CF = continuously flooded; (B) PF = periodically flooded; (C) MC = mesocosm control; (D) NC = nonmesocosm control; (E) FD = flooded–drained; and (F) +P and (G) +N, flooded–drained treatments with nutrients added. Lowercase letters indicate significant differences over time within a treatment (abc) (based on the method of least squares means with a Tukey adjustment [family error rate of P
Fig. 4. The effects of artificial flooding treatments on soil microbial P in field mesocosms. Brackets indicate periods of flooding. (A) CF = continuously flooded; (B) PF = periodically flooded; (C) MC = mesocosm control; (D) NC = nonmesocosm control; (E) FD = flooded–drained; and (F) +P and (G) +N, flooded–drained treatments with nutrients added. Lowercase letters indicate significant differences over time within a treatment (abc) (based on the method of least squares means with a Tukey adjustment [family error rate of P
Table 1. Physical and chemical characteristics of mesocosm soils collected in January 1992.
Values are sample means (n = 20) ± standard errors based on the standard deviation of n = 20 sample values divided by the square root of n = 20.
Loss on ignition.
Includes: 0.5 M NaHCO3 Pi and Po, microbial P, 0.1 M NaOH I and II Pi, and Po, 1.0 M HCl P, and residual P.
¶ Total P from digestion of nonextracted soils.
Table 2. Comparison of total P supply to in situ AER in treatment soils during flooded periods and nonflooded control soils during those same periods.
* Significant at the 0.05 level.
Values based on means of the total P supply (n = 4 per treatment) for each flooding duration.
Values based on means of the total P supply (n = 4 per control) for each respective time period.
FD = flooded–drained treatment.
¶ +P = flooded–drained treatment with P additions.
# N = flooded–drained treatment with N additions.
PF = periodically flooded treatment.
CF = continuously flooded treatment.
Table 3. Comparison of microbial biomass P in treatment soils during flooded months and control soils during those same months.
* Significant at the 0.05 level.
Values based on means of treatments (n = 4 per treatment) flooded during the respective month.
Values based on means of controls (n = 4 per control) during the respective month.
Jul = CF (continuously flooded), PF (periodically flooded treatment), FD (flooded–drained), +P (flooded–drained treatment with P additions), +N (flooded–drained treatment with N additions).
¶ Aug = CF, PF, FD, +P, +N.
# Sep = CF, FD, +P, +N.
Oct = CF, PF.
Nov = CF, PF.
Dec = CF.