Sediment phosphorus flux in an Oklahoma reservoir suggests reconsideration of watershed management planning

Abstract

A reservoir model of Lake Wister, Oklahoma, indicated that internal sources dominated phosphorus (P) loading to the waterbody, and that a watershed management plan need not address external P sources. To test this claim, we evaluated internal P loading by measuring sediment oxygen demand (SOD) and quantifying soluble reactive P (SRP) release from sediments to the overlying water column. Sediment cores from 3 sites were incubated at 21 C under quiescent conditions. Average SOD rates were between 9.9 and 22.6 mg/m²/h (240–540 mg/m²/d) across the three sites, where the shallow headwaters site had the least SOD. Average SRP release rate measurements varied from <0 to 3.30 mg/m²/d across the sites, and measurements were greatest under anaerobic conditions in the cores collected from deep water near the dam and the water intake structure. Measured values were an order of magnitude less than those used to calibrate the previous reservoir model. The new data contradict the findings of the earlier reservoir model and suggest that internal P cycling would not be the dominant P source. Thus, external P sources to Lake Wister must be considered in watershed management planning. A watershed-based strategy that focuses on both internal and external P sources is needed to improve overall water quality in this drinking water supply reservoir.

Key words:

• Lake Wister
• phosphorus release
• reservoir management
• sediment oxygen demand
• water quality

Pathogens, metals, nutrients, and sediments are generally the 4 main causes of beneficial use impairment in surface waters of the United States according to the Environmental Protection Agency's Total Maximum Daily Load Program assessment (USEPA 2009). Nutrients might represent the causal factor with the most complex relation to designated beneficial uses, including primary and secondary contact recreation, aquatic community health, and even drinking water supply (Dodds et al. 2009). Effects of elevated nutrients can be expressed at different trophic levels, causing shifts in algae, macroinvertebrates, and fish assemblages (Evans-White et al. 2009, Justus et al. 2010) in streams and rivers. There is, however, often a stronger correlation between nutrients and algae (expressed as chlorophyll a concentration) in lakes and reservoirs than in streams. In lakes and reservoirs, nutrients from external and internal sources supply the demand for phytoplankton growth. Phytoplankton sequester nutrients and inorganic carbon from the water column and eventually are deposited in the bottom sediments.

Figure 1 Lake Wister showing lake depths and sediment sampling sites where intact sediment cores were collected in June 2010 (adapted from PVIA 2009); the GPS coordinates were Site 1: 34°56′33.84^″N, −94°43′24.07^″W; Site 2: 34°56′45.83^″N, −94°43′12.48^″W; and Site 3: 34°55′10.31^″N, −94°47′20.79^″W.

External sources of nutrients include point-source effluent discharges from agricultural, industrial, and municipal facilities, as well as diffuse sources such as runoff from agricultural and urban landscapes. Nutrient loading to lakes and reservoirs, especially phosphorus (P) from external sources, is not only a concern with respect to phytoplankton production (Dillon and Rigler 1974, Canfield and Bachmann 1981), but may ultimately contribute to internal nutrient cycling. Some nutrients entering lakes and reservoirs may be adsorbed to particles in the water column and settle to the bottom (Sognozi et al. 1982, Correll 1998). Nutrients taken up by phytoplankton also reach the bottom when algal cells settle out. Nutrients in bottom sediments can be released back into the overlying water column through several processes, including reductive dissolution (Mortimer 1941), organic matter mineralization (Lovley and Phillips 1986, Andersen and Jensen 1992), and equilibrium concentration gradients (Haggard and Sharpley 2007, Belmont et al. 2009). In shallow areas, nutrient release into the water column can occur through wind resuspension, reductive dissolution during temporary anoxia under calm conditions, organic matter mineralization, and macrophyte senescence (Welch and Cooke 1995).

Reductive dissolution of iron (Fe) and manganese (Mn) minerals in sediments occurs when the overlying waters become anoxic and generally is considered the largest source of internal P release. The P bound to these minerals is released into the overlying waters when Fe and Mn are used as electron acceptors and reduced by bacteria. This phenomenon has been studied world-wide since the 1940s (Mortimer 1941, Cooke et al. 1977, Riley and Prepas 1984, Moore and Reddy 1994, Haggard et al. 2005) and shows that greater P release occurs with anaerobic conditions. Phosphorus release rates under such conditions vary from <1 mg/m²/d in oligotrophic to mesotrophic systems (Sen et al. 2007) to >15 mg/m²/d in highly eutrophic systems (Haggard and Soerens 2006). Sediment P release under aerobic conditions is probably regulated by mineralization rates (Andersen and Jensen 1992), Fe:P ratios (Jensen et al. 1992), and the equilibrium P concentration (Haggard and Soerens 2006, Spears et al. 2007) of bottom sediments. Some studies have suggested that P release from bottom sediments can be controlled by calcium minerals and can limit P release even under anaerobic conditions (Esten and Wagner 2010).

The overall goal of this project was to evaluate the P release from bottom sediments of Lake Wister, Oklahoma, under conditions of hypolimnetic anoxia. The specific objectives were to (1) measure the sediment oxygen (O₂) demand (SOD) of bottom sediments using flow-through cores, and (2) quantify P release rates from bottom sediments using static cores incubated under aerobic and anaerobic conditions. Phosphorus release rates from bottom sediments in regional reservoirs are typically less than loads from external sources (Haggard et al. 2005, Sen et al. 2007), providing <25% of the total annual P input to the water on average. Sediments, however, offer a management option in that P release to overlying water can be reduced by chemical treatment or oxygenation of the overlying waters (Kennedy and Cooke 1982, Welch and Schrieve 1994, Moore and Christensen 2009, Wauer et al. 2009, Huser et al. 2011).

Lake Wister has been listed on the Oklahoma Department of Environmental Quality (ODEQ) 303(d) list, which resulted in a landscape and reservoir modeling effort to identify P sources within its watershed (AMEC 2008). One of the conclusions based on these modeling efforts was that “with the internal load as the dominate pollutant source, this modeling project could not provide a reduction goal to draft a watershed management plan” (ODEQ 2010b). Nürnberg (2009) suggested that one of the concerns with reservoir modeling and internal loading was often inadequate determination of sediment P release. The purpose of this study was to fill a knowledge gap regarding the P release rates from bottom sediments of Lake Wister.

Table 1 Summary of designated beneficial uses, status, and cause per Oklahoma Department of Environmental Quality (ODEQ 2008, 2010a) and concentration ranges from recently available data (Patterson S, 2011 unpubl. data).

Beneficial Use	Status	Cause	Mean (Range)*
Public and Private Water Supply	Not Supporting	chl-a	22 (15–33) μg/L
Warm Water Aquatic Community (Fish & Wildlife Propagation)	Not Supporting	DO/Turbidity/pH	Depletion^1/20 (7–33) NTU/ 7.8 (7.2–8.5)^2
Aesthetics	Not Supporting	TP	0.07 (0.05–0.09) mg/L
Primary Body Contact Recreation	Not Supporting	Enterococcus	Not Reported^3
Fish Consumption	Not Assessed	Mercury	Not Reported
Agriculture	Supporting	—	—
*Data provided represent mean across sampling days and depths for the monitoring stations within Lake Wister, and range represents minimum and maximum mean; ^1Dissolved oxygen (DO) depletion generally occurs in the meta- and hypolimnion during stratified conditions; ^2pH was measured at 0.5 m below the surface; ^3Data were not reported for Enterococcus and mercury because these were not specific to this study nor measured by S. Patterson, 2011 unpublished data.

Study site description

Lake Wister is a drinking water supply reservoir in east-central Oklahoma with 2 primary inflows, the Poteau River and Fourche Maline Creek (Fig. 1). The US Army Corps of Engineers (USACE) began reservoir and dam construction in 1946, and its flood control operations were fully implemented in 1949. This reservoir has a mean depth of just over 2 m, a maximum depth of more than 15 m at normal conservation pool near the dam, and a surface area of approximately 2500 ha (OWRB 2003); the reservoir has a capacity of more than 62 million m³. The Poteau Valley Improvement Authority (PVIA) withdraws water (depth 1–2 m) from Lake Wister to provide drinking water to area residents (∼40,000 residents in LeFlore County, OK). Raw water is pulled from the 60 ha inlet or cove on the north-eastern corner of this impoundment for drinking water treatment.

Lake Wister is important not only for water supply, but for recreation, fishing, and water fowl hunting opportunities it provides for local residents and tourists. Sediment and nutrients entering Lake Wister from its watershed have affected the lake's water quality, reducing its recreational and water supply value. The reservoir was listed by the ODEQ for not supporting designated beneficial uses, including (1) public or private water supply, (2) warm-water aquatic community health, and (3) aesthetics (Table 1; ODEQ 2008). Lake Wister remained on the 2010 303(d) list as not supporting these designated beneficial uses, as well as primary body contact recreation (ODEQ 2010a). Since 1991, PVIA, the Oklahoma Water Resources Board (OWRB 2003), and USACE have been evaluating the nature of the water quality issues in the reservoir and working to implement projects to improve water quality. In late 2009, PVIA completed its strategic planning process for Lake Wister with the goal of meeting 4 objectives: (1) review issues critical to reservoir restoration and management, (2) review opportunities for restoration and enhancement, (3) develop a strategy to establish priorities and next steps, and (4) document discussions and decisions (PVIA 2009).

Methods and materials

In July 2009, 6 intact sediment–water interface cores were collected from 3 sites in Lake Wister for sediment-P release incubations. Cores were collected using a modified HYPOX corer (Gardner et al. 2009) and 0.5 m Plexiglas tubes with a 6.35 cm inside diameter. Tubes were inserted approximately 25 cm into the sediment using a weighted release mechanism on the coring device, and then the tubes were sealed at each end using #13 rubber stoppers. Site 1 was in the deepest part of the channel near the dam, and site 2 was in the deepest part of the cove behind Quarry Island and near the PVIA water intake structure (Fig. 1). Site 3 was in the headwaters at depths of 2 m or less.

Cores selected for the incubation study had a well-defined sediment-water interface with overlying water no more turbid than what was observed visually in the reservoir; thus, the chosen cores possessed sediment-water interfaces with minimal disturbance. Upon return to the lab, the water volume above the sediment in each core was reduced to 0.8 L. Next, each core was wrapped in aluminum foil to exclude light. Overlying water in all cores was aerated overnight, and thereafter cores were incubated at room temperature (∼21 C) under aerobic or anaerobic conditions. Three cores had air bubbled through the water column to maintain aerobic conditions, and the other 3 cores had nitrogen (N₂) gas with 270 ppm carbon dioxide (CO₂) bubbled through the water column to promote anaerobic conditions. After the first week, the cores incubated under anaerobic conditions were purged with N₂ gas immediately after sample collection and water replacement, and then the cores were sealed until the next sampling interval.

Water samples (∼0.05 L) were collected at 1 to 3 d intervals over the 10 d incubation, and removed water was replaced with reservoir water representative of the treatment conditions and with measured P concentrations. Water samples were filtered through a 0.5 μm glass-fiber filter, acidified to pH <2 using concentrated HCl, and analyzed for soluble reactive P (SRP). SRP was measured using the ascorbic acid reduction technique (EPA Method 365.2) at the Arkansas Water Resources Center Water Quality Lab, and other dissolved elemental concentrations (e.g., Fe and Mn) were measured using inductively coupled plasma optical emissions spectrometry (Spectro Genesis ICP–OES). Sediment release rates were estimated using the increase in P accumulated in the overlying water, which was corrected for SRP removed during sampling and added during water replacement. The amount of SRP released into the overlying water was normalized based upon initial SRP mass in the overlying water, and then the SRP mass (mg) released over time (d) was used to estimate sediment P flux (mg/m²/d) from the sediments across the surface area within the cores (0.0032 m²). Simple linear regression relating SRP mass released versus time was used to estimate sediment P flux from individual cores. Mean P flux was based on all site data under a given treatment (Haggard et al. 2005).

SOD measurements were made using flow-through intact cores (McCarty et al. 2007, Scott et al. 2008). Four intact sediment cores were collected at each site using the modified apparatus and 0.3 m Plexiglas tubes with a 10 cm inside diameter. Cores were capped with stoppers so that the overlying water volume was approximately 250 mL. Inflow tubing was passed through the stopper and positioned 2 cm above the sediment–water interface, and outflow tubing was flush with the bottom of the stopper at the upper end of the flow-through chamber. Inflow water was constantly aerated and then pumped into the cores at 0.03 L/h during the incubation at 21 C and in the dark. The inflow and outflow water was collected daily and then analyzed for dissolved O₂/argon (DO/Ar) ratios using a membrane inlet mass spectrometer (MIMS; Kana et al. 1994). The expected dissolved Ar concentration (mg/L) was calculated (based on Ar solubility at 21 C) and then multiplied by the (DO/Ar) ratio to get the DO concentration in each water sample. The difference in DO concentrations between the inflow (∼8.5 mg/L) and outflow was used to estimate the SOD (mg/m²/h), based on the flow rate (0.03 L/h) and sediment surface area within the cores (0.008 m²).

Table 2 Mean soluble reactive phosphorus (SRP) release rates or flux (mg/m²/d) from intact sediment cores collected on 14 July 2010 from 3 sites in Lake Wister, OK.

Site	Treatment	Slope (mg/d)	R^2	P Value	Mean Flux* (mg/m^2/d)
1	Aerobic	0.0024	0.41	0.17	0.75
1	Anaerobic	0.0048	0.78	0.02	1.52
2	Aerobic	0.0036	0.42	0.16	1.13
2	Anaerobic	0.0104	0.98	<0.01	3.30
3	Aerobic	0.0030	0.75	0.03	0.94
3	Anaerobic	−0.0007	0.42	0.16	−0.23
*Mean flux was reported as a numerical value even if the slope of the linear regression was not significantly different from zero (P > 0.05).

Results

Phosphorus concentrations in the overlying water

Initial SRP concentrations in the overlying water of the intact sediment cores varied after the water column was aerated overnight. Concentrations ranged from 0.025 to 0.162 mg/L in the overlying water on the first day samples (1 d). Initial SRP concentrations in the overlying water ranged from 0.025 to 0.075 mg/L in 12 cores, and the remaining 3 cores had concentrations >0.100 mg/L in the water column.

In general, SRP increased in the water column through the 10 day incubation in cores collected from all sites. SRP concentrations in the overlying water of an aerated core from Site 2 increased to a maximum of 0.182 mg/L over the 10 day incubation, whereas maximum SRP concentrations were generally <0.100 mg/L in all other aerated cores. Maximum SRP concentration in the overlying water of the anaerobic cores was 0.228 mg/L after 10 d, and >0.150 mg/L in all cores at the end of the incubation, except those from Site 3. SRP concentrations in the overlying water of the cores incubated under anaerobic conditions from Site 3 showed a slight decrease over time from 0.044 to 0.037 mg/L on average, in contrast to an increase observed at the site under aerobic conditions from 0.064 to 0.093 mg/L on average.

Phosphorus release rates from bottom sediments

The change in SRP mass over the first 24 h of incubation varied among treatments and sites, ranging from −0.003 to 0.021 mg SRP. Within 48 h, the change in SRP mass was positive in all cores, except those collected from Site 3 in the shallow headwaters. After 10 days, the average change in SRP mass was approximately 0.025 mg under aerobic conditions at all sites, whereas the change in SRP mass under anaerobic conditions averaged 0.049, 0.090, and −0.005 mg SRP on average for Sites 1, 2, and 3, respectively.

Slopes from the simple linear regressions of normalized SRP mass versus time were not significantly different from zero (P > 0.05) under aerobic conditions in the cores collected from deep water (Sites 1 and 2; Table 2; Fig. 2). There was variability between the cores at Site 2, where one core had SRP mass increasing over time but the other 2 showed no change. Incubation time explained 42% or less of the variability in mean SRP mass accumulated in the overlying water, and mean release rates were estimated to be 0.75 and 1.13 mg/m²/d from bottom sediments at Sites 1 and 2, respectively, under aerobic conditions. Individual cores from these sites had SRP release rates ranging from −0.43 to 4.20 mg/m²/d under aerobic conditions. The regression slope was significant (Site 3; 0.0030 mg/d, R² = 0.75, P = 0.03) in the aerated cores collected from the headwaters of Lake Wister (Fig. 2). Thus, average SRP mass release in these cores under aerobic conditions was 0.94 mg/m²/d.

SRP mass over time was less variable under anaerobic conditions compared to aerobic conditions between cores from individual sites. Regression slopes showed that significant SRP release occurred under anaerobic conditions in cores collected from deep water (Sites 1 and 2) and ranged from 1.52 to 3.30 mg/m²/d (Table 2; Fig. 3). SRP release rates were greatest in the cores taken near the raw water intake (Site 2), ranging from 2.95 to 3.75 mg/m²/d under anaerobic conditions. In contrast, SRP release rates near the dam (Site 1) were nearly identical across the cores under anaerobic conditions (1.51–1.52 mg/m²/d). The slope of mean SRP mass accumulated over time in the overlying water of the cores from Site 3 was not significantly different from zero (−0.0007 mg/d, R² = 0.42, P = 0.16) under anaerobic conditions (Fig. 2).

SRP mass released from the bottom sediments was positively correlated to the dissolved Fe concentrations in the overlying water in the cores (r = 0.52, P = 0.001). Dissolved Fe concentrations at the beginning of the incubations were ∼1 mg/L in the overlying water of the sediment cores across all sites, whereas initial dissolved Mn concentrations in the overlying water of the cores varied between the deep water sites (2–4 mg/L) and the shallow site (∼1 mg/L). Dissolved Mn concentrations were also positively correlated to SRP mass released from the bottom sediments in these incubations (r = 0.33, P = 0.049).

Sediment oxygen demand

Flow-through water used to estimate potential SOD had an initial DO concentration of 8.5 mg/L for the cores collected from Sites 1 and 2, and 8.4 mg/L for those collected from Site 3. DO concentrations in the outflow water varied among the cores collected at deep sites (Sites 1 and 2) and those collected at the shallow site in the headwaters (Site 3). The outflow from the cores collected at Sites 1 and 2 had DO concentrations ranging from 1.1 to 4.1 mg/L, whereas the outflow DO concentration from Site 3 cores was between 5.4 and 5.8 mg/L. Differences in outflow DO concentrations translated into differences in potential SOD (Table 3), where SOD varied from a mean of 22.5 mg/m²/h from the bottom sediments in deep water to 9.9 mg/m³/h on average from the bottom sediments in the shallow headwaters.

Figure 2 Soluble reactive phosphorus (SRP) mass released into the water column of the sediment cores under aerobic conditions from Lake Wister, OK, as a function of time; the data represent the average (± standard deviation) from the 3 cores, and the regression lines are provided to visualize the general relation.

Figure 3 Soluble reactive phosphorus (SRP) mass released into the water column of the sediment cores under anaerobic conditions from Lake Wister, OK, as a function of time; the data represent the average (± standard deviation) from the 3 cores, and the regression lines are provided to visualize the general relation.

Discussion

Complex reservoir models rely on site specific measurements to reduce uncertainty in model output, and when data are not available, model parameters are often used in the calibration process. The reservoir model of Lake Wister included sediment–water P interactions, and this parameter, representing P flux from bottom sediment, was used in the calibration. The P release rates used in model parameterization were not measured but were assumed to represent realistic values for this reservoir. Model P release rates were 30 mg/m²/d in zone 1, representing the flux in deep water (i.e., near the dam and the PVIA raw water intake structure), and 50 mg/m²/d in zone 2, representing the flux from shallow waters (e.g., headwaters and shorelines; AMEC 2008). The SOD rates used in the model were approximately 150 and 200 mg/m²/h in Zones 1 and 2, respectively. SOD, although also unmeasured, was used in model calibration to fit observed DO concentration profiles in Lake Wister. The largest uncertainty in this reservoir model was likely in these two calibration parameters because site specific measurements were not available. This study measured P release rates from and O₂ demand in bottom sediments of Lake Wister, and the measured rates were different from those used to parameterize the reservoir model. Potential SOD measured from the collected cores (9.9–27.6 mg/m²/h) was an order of magnitude less than that used in model parameterization (150–200 mg/m²/h; AMEC 2008). Potential SOD rates were quantified at only 3 sites at Lake Wister, which might not capture the range in SOD across this reservoir, but these measurements were within the range observed in other regional reservoirs. For comparison, estimated potential SOD rates at Beaver Lake, Arkansas, were 8.8–28.5 mg/m²/h using this same method (Hamdan et al. 2010). There are several ways to measure potential SOD rates of bottom sediments including laboratory and in situ methods; lab methods are generally more consistent, reproducible, and efficient than in situ methods (Bowman and Delfino 1980). This study used lab methods, which produced repeatable SOD results from 4 replicates at each site. These SOD rates were measured under quiescent conditions in the lab, and turbulent conditions can increase SOD up to 5-fold (Beutel et al. 2007). Even with this induced oxygen demand, SOD rates were still less than those used in the model. Overall, SOD is an important component of reservoir sediment dynamics and modeling because it drives the reducing conditions promoting P release from bottom sediments.

Table 3 Potential sediment oxygen demand (SOD) rates or flux (mg/m²/h) from intact sediment cores collected on 14 July 2010 from 3 sites in Lake Wister, OK.

Site	Observations	Mean SOD (mg/m^2/h)	SOD Range (mg/m^2/h)
1	4	22.6	18.1–27.6
2	4	22.5	16.2–26.8
3	4	9.9	9.4–11.2

Table 4 Estimated soluble reactive phosphorus (SRP) release rates or flux (mg/m²/d) from intact sediment cores collected during summer at other regional reservoirs near Lake Wister.

Reservoir	Site Description	Aerobic Flux (mg/m^2/d)	Anaerobic Flux (mg/m^2/d)	Reference
Lake Wister, Oklahoma	Shallow Waters	50 (calibrated parameter)		AMEC (2008)
	Deep Waters	30 (calibrated parameter)
Beaver Lake, Arkansas	Riverine Zone	0.13	0.85	Sen et al. (2007)
	Transition Zone	0.15	1.77
	Lacustrine Zone	0.04	<0.01
Lake Eucha, Oklahoma	Riverine Zone	1.14	4.70	Haggard et al. (2005)
	Transition Zone	1.01	2.46
	Transition Zone	0.95	6.05
Lake Frances, Arkansas – Oklahoma	Headwaters	0.37	14.53	Haggard and Soerens (2006)
	Near Dam	3.57	15.53

The P release rates used in model parameterization for Lake Wister (30–50 mg/m²/d) were much greater than those measured in this study using intact sediment cores (∼0–3.30 mg/m²/d) and reported throughout the literature. Sediment P release rates used in the reservoir model (AMEC 2008) were near past measurements for Lake Mendota (Holdren and Armstrong 1980) that were an order of magnitude greater than other reported values. SRP release rates measured at Lake Wister were within the range reported for other regional reservoirs (<0.01–16 mg/m²/d), which were considered to express mesotrophic to eutrophic conditions (Table 4). Lake Wister might be best compared to Lake Eucha, Oklahoma, where SRP release rates from bottom sediments were estimated to be ∼1 mg/m²/d under aerobic conditions and from 2.46 to 6.05 mg/m²/d under anaerobic conditions (Table 4; Haggard et al. 2005). The measured P release rates at Lake Wister were an order of magnitude less than the value used in the reservoir model calibration, suggesting that the reservoir model should be recalibrated with these measured values. The difference between the measured rates and those used in model calibration does emphasize that complex reservoir models are enhanced by site-specific data. It is possible that model (AMEC 2008) parameterization overestimated settling within this reservoir, compensating for the elevated internal loading used in the model. Internal P loading might not be the dominant source to Lake Wister, and catchment P sources to Lake Wister should not be neglected when developing watershed-based strategies to improve water quality.

Although this study estimated sediment P release in cores from only 3 sites within Lake Wister, SRP concentrations in the overlying water of the cores and past measured concentrations above the sediment–water interface in this reservoir were comparable. The reservoir model was parameterized to match data available from the 2001 study, in which total P (TP) concentrations were available from the surface and bottom of the water column at sites near the dam and PVIA water intake structure (AMEC 2008, ODEQ 2010b). In summer 2001, the observed TP concentrations near the reservoir bottom increased to ∼0.13 mg/L near the dam and to ∼0.27 mg/L near Quarry Island where the PVIA water intake structure is located. Overlying water in the sediment cores collected from these sites had a mean SRP concentration of 0.16 mg/L under anaerobic conditions during the last few days of the incubations. Concentrations in the overlying water were fairly similar to that observed in Lake Wister, especially considering that SRP only makes up a portion of TP. The overlying water in the cores was relatively clear at the end of the incubation, suggesting P was present mostly in the dissolved form.

Phosphorus can be released from sediments to the overlying water in the dissolved reactive form, as measured in this study, or as dissolved organic P (DOP), which was not measured in this study. Hamdan et al. (2010) observed that DOP releases were up to 5 times greater than SRP in sediment cores collected from deep water at Beaver Lake, Arkansas. SRP release rates at Beaver Lake, however, were generally less than those observed at Lake Wister (Table 4; Sen et al. 2007, Hamdan et al. 2010). Wang et al. (2009) found that DOP release was the largest single fraction of total dissolved P release from sediments, representing 30–60% of the P released from dried sediments into aqueous solution under aerobic conditions. DOP release could possibly be important at Lake Wister and should be the focus of future sediment studies. Most studies, however, have shown that DOP release rates are of the same order of magnitude of SRP release, suggesting that the values used in the reservoir model still overestimated the P released from the bottom sediments into the overlying water at Lake Wister.

Phosphorus can also be resuspended from bottom sediments under turbulent mixing conditions, especially in shallow headwaters. Wind resuspension of unconsolidated bottom sediments might increase TP concentrations in the water column of shallow lakes and in shallow waters (Søndergaard et al. 1992, Welch and Cooke 1995). This P, however, is usually bound in particulate forms that are less bioavailable relative to SRP released from bottom sediments. Equilibrium P concentrations (EPC₀) of the suspended and bottom sediments would probably regulate dissolved P in the overlying waters, and Spears et al. (2007) showed that EPC₀ was less in shaken (i.e., resuspended) sediments (0.03 mg/L) than quiescent, intact sediment cores (0.18 mg/L). Sediment resuspension occurs in Lake Wister, although the bioavailability and EPC₀ have not been evaluated in this reservoir. The release of SRP from resuspended sediments at Lake Wister is likely similar to aerobic release rates (1 mg/m²/d) measured in this study.

Hypolimnetic O₂ and sediment P release are potential management foci in reservoirs. For example, Moore and Christensen (2009) showed that water column oxygenation coupled with an alum injection system to reduce P release could yield improved water quality. Furthermore, chemical treatment (e.g., alum) of reservoir bottom sediments could reduce P flux by 85% in the first couple years (Huser et al. 2011) as well as P concentrations in the water column (Wauer et al. 2009, Huser et al. 2011). Hypolimnetic oxygenation and chemical treatment (e.g., alum injection) might be a potential management option for Lake Wister, especially for the portion of the reservoir near the PVIA water intake. Such in-lake approaches to reservoir management must, however, be coupled with watershed management strategies that reduce P inputs from external sources. James and Pollman (2011) suggested that coupled internal and external P management would decrease the time required to meet water-quality goals, but that the economics of internal P management should also be considered.

Summary

We measured SOD and P release rates under quiescent conditions in intact sediment cores collected at Lake Wister in July 2010, and then compared these measured values to those used in calibration of the previous reservoir model (AMEC 2008). The mean O₂ demand by bottom sediments was from 9.9 to 22.6 mg/m²/h on average across the sites, and the least SOD was measured in the shallow reservoir headwaters. Measured rates were an order of magnitude less than those used in the reservoir model parameterization, suggesting that reservoir models should incorporate empirical data and consider SOD under both quiescent and turbulent conditions. Oxygen demand can drive hypolimnetic waters to anoxia with the potential release of SRP from the bottom sediments through reductive dissolution and microbial mineralization. Measured SRP release rates (∼0–3.30 mg/m²/d) were also an order of magnitude less than those used in the reservoir model calibration. The previous modeling report (AMEC 2008) suggested that internal P sources were dominant in the reservoir nutrient budget, and that this large source precluded the development of a watershed-based strategy to reduce P inputs from external sources. This study illustrates the importance of acquiring empirical measures of SOD and SRP release rates to parameterize reservoir models, rather than relying on what might appear to be reasonable parameter estimates as calibration tools.

Acknowledgments

We thank M. Bonaduce, J. Corral, and B. Baker for assistance in the collection of sediment cores, incubations and sampling of the overlying water during the incubations. Funding for this study was provided by the Poteau Valley Improvement Authority. This manuscript benefited by reviews, comments and suggestions from Editor Dr. K.J. Wagner, Associate Editor Dr. R.T. James, and 2 anonymous reviewers. We appreciate the patience and assistance of the journal editors.