Effectiveness of alum in a hypereutrophic lake with substantial external loading

ABSTRACT

Brattebo SK, Welch EB, Gibbons HL, Burghdoff MK, Williams GN, Oden JL. 2017. Effectiveness of alum in a hypereutrophic lake with substantial external loading. Lake Reserv Manage. 33:108–118.

Water quality conditions in hypereutrophic Lake Ketchum improved to mesotrophic conditions after alum treatments in 2014 and 2015. From 2013 pre-treatment conditions, summer mean epilimnetic total phosphorus (TP) declined from 289 to 34 µg/L in 2014 and then to 15 µg/L in 2015 (a total reduction of 95%). Hypolimnetic TP declined 99% overall, and chlorophyll a dropped 70% while there was 135% improvement in transparency over the 2 years. Toxic cyanobacteria blooms, a chronic problem in the lake, disappeared in 2015. Maintenance treatment in 2016 continued water quality improvements. The primary driver of water quality changes was a decrease in internal P loading from a mean rate of 25 mg/m² per day to zero (negative rate) after the alum treatments. The improvements occurred despite application problems with the 2014 treatment, which had to be halted with only 66% of the planned dose (aluminum: 28 mg/L or 83.7 g/m²) applied. The remaining 34% of the original alum dose plus an annual maintenance water column dose and an additional 15% was applied in March 2015. The water quality improvements were achieved even with continuation of a highly enriched (TP: >450 µg/L) external input from a legacy agricultural source. The changes were consistent with or exceeded predictions of a 2-layer, seasonal mass balance TP model used to evaluate restoration alternatives. These results show that alum treatments can eliminate high internal loading and, when coupled with annual maintenance treatments, provide sustained water quality improvements despite continued external loading.

KEYWORDS:

• Alum
• external loading
• hypereutrophic lake
• internal loading

Aluminum (Al) salts, in particular Al sulfate (alum), have been a commonly used and effective in-lake technique to improve water quality in eutrophic lakes (Huser et al. 2015a, 2015b). There have been at least 250 alum treatments worldwide since the first in Sweden in 1967 (Cooke et al. 2005, Brattebo et al. 2015). The first treatments in the United States were Horseshoe and Snake lakes, Wisconsin, in the early 1970s (Huser et al. 2015a). While long-term external loading is the primary source for phosphorus (P) enrichment in lake sediments, the release of that stored P from lake sediments is often the principal cause of summer algal blooms (Welch and Jacoby 2001). In many cases, control of the internal source may be more effective at improving lake quality than controlling just external sources, especially if implementation of watershed best management practices (BMPs) and other control techniques have been exhausted or are too costly. For example, alum was found to be 50 times more effective in terms of P removed per dollar spent than catchment BMP measures in one group of urban lakes (Huser et al. 2015b). Alum treatments, if dosed properly and applied correctly, have reduced P loading from sediments by 70–85% on average, which in turn has reduced lake total P (TP) and decreased algal bloom frequency and density (Cooke et al. 2005, Huser et al. 2015b). Also, alum can provide long-term effectiveness at reducing P and controlling nuisance algal blooms (Brattebo et al. 2015). The longevity of 114 alum treatments averaged 15 years for lakes with water quality improvement, such as decreased epilimnetic TP, increased water clarity, and decreased phytoplankton biomass (Huser et al. 2015a).

Lake Ketchum, in the northwest United States, was hypereutrophic, with a summer mean epilimnetic TP concentration of 289 µg/L and frequent toxic algal blooms. The purpose of our project at Lake Ketchum was to design and implement a program of alum treatments to control internal P loading and reduce in-lake P concentrations enough to limit algal blooms and curb toxic algae incidents. The questions were: would inactivation of sediment P with alum treatments in a strongly stratified lake reduce summer epilimnetic TP concentrations enough to provide water quality improvements, and would continued external P loading from a former agricultural area offset the benefits of the alum treatments? We used a 2-layer, seasonal mass balance TP model to select the best restoration alternative and to simulate lake response to the proposed alum treatments.

Monitoring data from 2010 to 2011 and the calibrated TP model showed that internal loading contributed 73% of the annual P load (Table 1). The magnitude of the sediment P release rate (SRR) before alum treatments was extremely high, averaging 25 mg/m² per day and reaching a maximum of 42 mg/m² per day. Typical release rates of P for eutrophic lakes stratified with anoxic conditions are in the range of 2 to 5 mg/m² per day (Welch and Cooke 1995). Late summer hypolimnetic TP reached concentrations of 3600 µg/L in 2011 and 3200 µg/L in 2013, approaching the level of untreated sewage. Even if reduced by 85%, a commonly observed effectiveness of alum treatments (Huser et al. 2015b), there would still be high SRR and high P concentrations in the hypolimnion that might be available for algal production through diffusion, entrainment, and mixing.

Table 1. Total phosphorus loading to Lake Ketchum, October 2010–October 2011.

Source	Total phosphorus inflows (kg)	% Phosphorus load
Internal (sediments)	455	73.0%
Inlet stream	146	23.4%
Surface runoff	13	2.1%
Groundwater (GW)	7	1.2%
Direct precipitation	2	0.3%
Total	623 (6414 mg/m^2 per year)	100%

In addition, the main lake inlet stream, which drained from a former agricultural area, still contained an average TP concentration of 474 µg/L (85% soluble reactive P [SRP]) and contributed 23% of TP load to the lake (Table 1). A contention remains among some investigators that alum should not be used as an in-lake restoration tool until external sources are eliminated or greatly reduced. One purpose of this project was to determine if alum treatments could be successful even with continued high external loading.

Study site

Lake Ketchum is a small lake (10.5 ha) located in Snohomish County, north of Seattle, Washington. The lake is relatively shallow (maximum depth 6.7 m), but stratifies strongly from April to October. The lake's Osgood Index (mean depth/√surface area) of 12 represents a high resistance to mixing (Osgood 1988). The primary lake inflow is from a small stream that enters on the southeast corner. A wetland on the northeast shore, several small ditches, and groundwater also feed the lake. A seasonal outlet drains to the west to Skagit Bay. The watershed covers ∼42 ha, mainly comprising semi-rural residential, with higher density suburban residential development immediately surrounding the shoreline (Burghdoff and Williams 2012).

A former dairy farm covers ∼8 ha at the south end of the watershed. This farm also served as an animal waste depository for several years, resulting in high soil P concentrations, ranging from 300 to 650 mg/kg (Bray P soil test) over much of the farm (Snohomish Conservation District, October 1999, unpubl. data). Since 1996, several improvements to the runoff from the former dairy farm have reduced P entering the lake (Burghdoff and Williams 2012). Cattle were removed from the portion of the farm draining to the lake, and the land was converted to hay, which was cut and removed each year. In addition, the import and storage of animal waste ceased. These improvements led to significant reductions in lake inflow TP concentrations; average winter inflow TP concentrations declined from nearly 1500 µg/L during 1995–1996 to 557 µg/L in 2011 (Burghdoff and Williams 2012). However, further actions that would reduce P export from the farm are not feasible at this time.

Prior to the alum treatments, the lake was considered the most hypereutrophic lake in Snohomish County. Algal scums covered the entire lake surface for most of the summer. The lake suffered from chronic, toxic cyanobacteria blooms, with microcystin concentrations reaching 551 µg/L in 2011, 417 µg/L in 2012, and 539 µg/L in 2013. During these years, the lake was posted for toxic blooms over 19 weeks per year on average. Recreational activities and other beneficial uses were severely limited.

Methods

Water quality

Lake water quality monitoring was conducted at a deep mid-lake station. To develop the 2-layer mass balance model, samples were collected once a month from October 2010 through March 2011 and twice a month from April through October 2011. Samples were analyzed for TP and SRP at 1, 2, 4, and 5 m depths and for chlorophyll a (Chl-a) at 1 m. The June through September samples from 1 m and 5 m were also analyzed for total Kjeldahl nitrogen (TKN) and nitrate + nitrite (NO₃ + NO₂) nitrogen. At each sampling event, temperature and dissolved oxygen (DO) were measured in situ at 1 m intervals throughout the water column using a Hydrolab MiniSonde MS4A or a YSI Pro 20 Temp/DO meter. Transparency was determined with a Secchi disk. Samples were also collected for TP and SRP from the inlet stream (when flowing) on the same dates as the in-lake sampling.

To evaluate pre- and post-treatment conditions, samples were collected monthly beginning in March 2013 and twice a month during the stratified periods in 2014 and 2015. In 2016, samples were collected once a month. Samples were analyzed for TP and SRP at 1 and 5 m depths, for Chl-a at 1 m, and for TP and SRP at the inlet stream. Samples were collected for total persulfate nitrogen (TN) from 1 m beginning in November 2013 and occasionally from 5 m. Temperature, DO profiles, and transparency were measured during each sampling event.

All water samples were analyzed following standard methods (Eaton et al. 2005) by IEH–Aquatic Research, Inc. in Seattle, Washington. TP and SRP were determined using SM 18 4500 PF with a method detection limit (MDL) of 2 µg/L for TP and 1 µg/L for SRP. Chl-a was determined using SM 18 10200 H with a MDL of 1 µg/L. TKN was determined using EPA 351.1 with a MDL of 200 µg/L, and NO₃ + NO₂ was determined using SM 18 4500NO3F with a MDL of 10 µg/L. TN was determined using SM 20 4500 NC with a MDL of 50 µg/L.

Sediment sampling

Three sediment samples (one in shallow water, one in the deep mid-lake station, and one in a deep station replicate) were collected in 2011 using a piston interface corer (Aquatic Research Instruments 2011) following common sediment core sample procedures (Cushing and Wright 1965, Wright 1980, 1991, Blomqvist 1991, Fisher et al. 1992). Sediment samples were also analyzed by IEH–Aquatic Research Inc. Core samples were sectioned according to the Washington Department of Ecology Standard Operating Procedure 038 and Rydin and Welch (1998). Each section core was fractionated for water content (using EPA Method 365.1) and then P content to assess bound P (Al, iron [Fe], calcium) and organic P using reagents described by Rydin and Welch (1998) and Casbeer (2009).

TP mass balance model

A 2-layer, seasonal mass balance TP model (equation 1(1) ) was developed to plan and evaluate lake treatment alternatives. Using a 2-week time step, the model was calibrated against the observed Lake Ketchum data from October 2010 through October 2011. Development and calibration followed Perkins et al. (1997) and Auer et al. (1997). The model used the external inputs and lake outputs from the P budget developed for the lake (technical appendix in Burghdoff and Williams 2012; Table 1) as the basis for calculating and calibrating the internal P cycling, including the rates of diffusion, entrainment, sedimentation, and sediment release: (1)

Diffusion of P from high hypolimnetic concentrations to lower epilimnetic concentrations was accounted for using temperature gradients between stratified layers and the vertical heat exchange coefficient. An exchange coefficient of 0.054 m per week or 0.109 m for every 2 weeks was applied across the entire stratified period. This low rate of diffusion corresponds to the high resistance to mixing reflected in the lake's Osgood Index value of 12 (Cooke et al. 2005). Temperature profiles and observed TP indicated that no direct physical entrainment of hypolimnetic water into the epilimnion was observed during summer 2011; therefore, the modeled thermocline midpoint was held constant at 2.5 m. During the last time step (11 Oct 2011 to 24 Oct 2011), when turnover occurred, entrainment was observed and the thermocline depth was increased to 3 m.

Settling rates during the non-stratified period (Oct–Apr) were determined by calibrating against the observed whole-lake TP concentration (Fig. 1). The rate ranged from 0.03 m per week at the beginning of the period to 0.20 m per week at the end. As the lake stratified, settling from the epilimnion began at 0.5 m per week and gradually increased to 0.7 m per week at turnover in late September. Hypolimnetic SRP concentrations were nearly the same as TP concentrations throughout the stratified period. The settling rate in the hypolimnion was assumed to be zero because P was released (diffused) from sediment pore water as SRP, which did not change until turnover when particulate P was formed with Fe.

Figure 1. Modeled vs. observed whole-lake volume weighted TP concentrations in Lake Ketchum, WA, Oct 2010–Oct 2011.

The gross SRR was calculated from the change in hypolimnetic TP concentration during the stratified period. No settling loss occurred, as indicated by the nearly equal TP and SRP concentrations (Fig. 2). Average SRR was estimated by regression analysis at 32 mg/m² per day during the anoxic stratified period in 2011, with a maximum of 42 mg/m² per day. TP content reached a high of 3600 µg/L in the hypolimnion. To calibrate the model, the SRR was increased to 45 mg/m² per day during the stratified period to compensate for the probable loss of P with groundwater. An estimated 0.5 kg per year, or 1370 mg per day, of P was lost to groundwater in the model.

Figure 2. Observed phosphorus concentrations at 1 m, 5 m and in the main inlet of Lake Ketchum, WA, from Oct 2010 to Oct 2011.

We assumed that aerobic sediment release from epilimnetic sediments occurred during the study period. Shallow, unstratified aerobic lakes have summer internal loading (e.g., 6 mg/m² per day in Upper Klamath, Oregon, and 2.6 mg/m² per day in Long Lake [Kitsap], Washington; Welch and Cooke 1995). Aerobic release is especially likely if the total Fe (TFe) to TP ratio in surficial sediment is <15:1 (Jensen et al. 1992), which was the case in Lake Ketchum, which ranged from 7:1 to 14:1. A rate of 2 mg/m² per day for epilimnetic sediment aerobic release was assumed during the stratified period in Lake Ketchum.

Several iterations and adjustments were required during calibration. For the nonstratified period, known ranges of settling were used to compare the model output to the observed lake conditions. The observed hypolimnetic SRR during the stratified period was used to conform to the observed lake TP, and the settling rate was adjusted accordingly. The final calibrated model closely aligned with the observed lake values (Fig. 1).

Aluminum dose

The required Al dose to inactivate sediment P was calculated based on the P fractions analyzed from the sediment cores. The available P (loosely sorbed-P + Fe-P + biogenic-P) in the sediments averaged 0.805 mg/g in the top 10 cm (Burghdoff and Williams 2012). A ratio of 20:1 for Al added to available P was selected to determine the dose. That ratio over 10 cm was similar to the 100:1 ratio for mobile-P only over 4 cm described in Rydin and Welch (1999). The calculated Al dose to treat all the lake's sediments was 83.7 g/m² or 28 mg/L, including 4 mg/L for TP in the water column. This high dose required application of both alum (8.0–8.4% Al₂O₃) and a buffer of sodium aluminate (32–38% soluble NaAlO₂). Alum and sodium aluminate are often used jointly in low alkalinity lakes, such as Lake Ketchum, to mitigate impacts to pH. Although sodium aluminate was used, the treatment was referred to as an alum treatment. At a volume ratio 2:1, the planned application required 77,162 L of alum and 38,581 L of sodium aluminate applied to the whole lake. The planned application was to cover the whole lake area, adjusting dose of each chemical given depth of the lake to maintain 28 mg/L of Al.

Annual spring water column maintenance alum treatments were planned for subsequent years following the whole-lake sediment inactivation. The recommended annual Al dose was 4.1 mg/L based on historic water column TP concentrations following turnover; however, it was also recommended to Snohomish County that any future water column maintenance dose be adjusted based on pre-stratification water column TP concentrations, especially in wet years with higher external TP inputs.

Results

Model analysis of alum treatment alternatives

Four alum treatment alternatives were evaluated with the calibrated TP model. These alternatives were (1(1) ) whole-lake sediment inactivation with alum, (2) whole-lake sediment inactivation repeated in 2 years, (3) whole-lake sediment inactivation with alum injection to the inlet stream, and (4) whole-lake sediment inactivation with annual water column alum treatments. Each of the alternatives was aimed at reducing the high internal P loading from the lake sediments that had resulted from long-term inputs from agricultural activities in the watershed. Alternatives 3 and 4 also addressed the continued high external P loading from the legacy of past agricultural practices. The goal of annual water column alum treatments was to inactivate the continued high external P loading annually, which mainly occurs during the winter and spring months.

The lake water quality goal set by Snohomish County was to maintain the summer (May–Oct) epilimnetic TP concentration at or below 40 µg/L. Environmental conditions in the treatment years were assumed to be similar to those during the study year (Oct 2010 to Oct 2011). An 85% reduction in internal loading of P from both anoxic and oxic sediments (Huser et al. 2015b) was assumed as the overall treatment effectiveness.

The model predicted lake TP responses for each restoration alternative (Table 2).^{Table 3}Alternative 1 would not meet the goal of 40 µg/L for TP in the epilimnion, and neither would Alternative 2, except for year 3 following implementation of the second whole-lake treatment, after which mean summer TP would increase again. The model predicted that both Alternatives 3 and 4 would lower TP to meet the goal beginning in year 2, when either the stream alum injection system or the annual water column alum treatments would begin controlling the external P load coming from the inlet stream. Alternative 3, however, would provide slightly better water quality (Table 2^{Table 3}).

Table 2. Summer (May–Oct) mean epilimnion TP concentrations (µg/L) as predicted by a mass balance model for various lake treatment alternatives in Lake Ketchum.

Treatment scenario	Pre	Year 1	Year 2	Year 3	Year 4
Observed	181	—	—	—	—
Modeled existing conditions	—	175	177	178	178
Alternative 1: Whole-lake sediment inactivation alum treatment	—	53	79	80	80
Alternative 2: Whole-lake sediment inactivation alum treatment repeated in year 3	—	53	79	38	64
Alternative 3: Whole-lake sediment inactivation alum treatment with alum injection on inlet stream	—	53	37	37	37
Alternative 4: Whole-lake sediment inactivation alum treatment with annual water column alum treatments	—	46	40	40	40

The alternatives were then evaluated based on estimated annual, 4-year, and 10-year costs (Table 3). Although Alternative 3 was the option most likely to meet the water quality goal, it was also the most expensive. Alternative 4 was considerably less expensive than Alternative 3 over the first 4 years, but because of the uncertainty of the long-term effectiveness of annual water column treatments in controlling the continued external loading from past agricultural activities, the whole-lake sediment alum treatment might need to be repeated within 5–7 years. This assumption factored into the 10-year cost estimate for Alternative 4.

Table 3. Cost comparison of treatment alternatives for Lake Ketchum evaluated by the mass balance model (Burghdoff and Williams 2012).

	Cost in $1000s (2011 dollars)
Lake restoration alternative	Year 1	Year 2	Year 3	Year 4	4-Year Total	10-Year Total
Alternative 1: Whole-lake sediment inactivation alum treatment	$194	0	0	0	$194	$194^a
Alternative 2: Whole-lake sediment inactivation alum treatment repeated in year 3	$194	0	$174	0	$368	$890^b
Alternative 3: Whole-lake sediment inactivation alum treatment with alum injection on inlet stream	$428	$43	$65	$65	$601	$991^c
Alternative 4: Whole-lake sediment inactivation alum treatment with annual water column alum treatments	$230	$36	$36	$36	$338	$692^d

^a No repeated treatments in first 10 years.

^b Sediment treatments repeated every 2 years.

^c Sediment treatment will not need to be repeated until after 10 years.

^d Sediment treatment will need to be repeated in Year 6.

The final recommended in-lake alternative, based on model predictions and costs, was Alternative 4 (whole-lake sediment inactivation alum treatment with annual water column treatments) because it minimized cost and provided the most flexibility.

Alum treatments

The first of 2 alum treatments occurred at Lake Ketchum on 20 and 21 May 2014. Instead of the planned 2:1 ratio (77,162 L of alum and 38,581 L of sodium aluminate) required to provide an Al dose of 28 mg/L, the chemical ratio was revised to 1.8:1 and the volume of the sodium aluminate buffer was increased to 42,824 L when on-site jar tests determined that the strength of the delivered buffer was weaker than originally specified.

Unfortunately, the treatment was stopped prior to completion because of multiple operational problems that resulted in adverse environmental conditions: large pH swings, incomplete mixing of chemicals, inadequate floc formation, and fish mortality. Only 66% of the total dose was applied to the lake in 2014 (51,042 L alum and 28,069 L of sodium aluminate).

On 4 and 5 March 2015, the remainder (34%) of the originally planned dose plus the annual water column stripping dose planned for 2015 and an additional 15% of the original dose was applied to the lake. The application ratio was further revised to 1.6:1 during the 2015 treatment (vs. 1.8:1 in 2014) based on on-site jar tests and delivered buffer strength. The total volumes applied in 2015 were 49,210 L of alum and 30,730 L of sodium aluminate. The alum applicator made adjustments to the application method and completed the 2015 treatment without operational or environmental complications.

In 2016,^{Figure 3}the planned annual water column maintenance treatment was applied to the whole lake. Total volumes of chemical applied in 2016 were 10,978 L of alum and 6,454 L of sodium aluminate. The application ratio for 2016 was 1.7:1 based on jar tests and delivered buffer^{Table 4Figure 3}strength.

Lake water quality

Phosphorus and nitrogen

TP concentrations in Lake Ketchum during 2011 and 2013, prior to treatment, were extremely high (Table 4, Figs. 2–4). During 2011, summer (May–Oct) average TP at 1 m and 5 m depths was 181 µg/L and 2667 µg/L, respectively. In 2013, summer mean epilimnetic TP was higher at 289 µg/L, whereas hypolimnetic mean TP was less but still very high at 1427 µg/L (Table 4). Hypolimnetic P was mostly soluble as SRP, 81% in 2011 and 87% in 2013, indicating that internal loading of P from the sediments was likely the main source confirming the P budget. During October 2010 through October 2011, internal loading accounted for 73% of the TP load to Lake Ketchum (technical appendix in Burghdoff and Williams 2012; Table 1). During the summer months (May–Oct), internal loading contributed 98% of the TP load in 2011. Peak TP in the hypolimnion reached 3620 µg/L in 2011 and 3260 µg/L in 2013. Sediment release rates (based on regression analysis) ranged from 32 to 42 mg/m² per day during 2011, 12.5 to 21.6 mg/m² per day in 2013, and 19.7 mg/m² per day in early spring 2014 prior to the first alum treatment.

Table 4. Summer (May–Oct) mean nutrient and trophic state indicators before and after 2 successive alum treatments (µg/L) in Lake Ketchum. Inlet stream nutrient concentrations are annual means.

	Pre-treatment		Post treatment
Parameter	2011	2013	2014	2015	2016
TP – 1 m	181	289	34	15	16
TP – 5 m	2667	1427	186	17	23
SRP – 1 m	29	—	3	2	1
SRP – 5 m	2165	1235	83	1	2
TN – 1 m	1240^a	—	1166	663	552
TN – 5 m	4620^a	—	1639	698	752
TN:TP ratio	6.9	—	34.1	43.8	35.6
Chl – 1 m	81	56	45	17	4
Microcystin	0.94–551	0.54–539	16.1–675	n/a	n/a
Secchi (m)	1.0	1.7	2.1	4.0	3.5
TP - Inlet	557	474	458	456	385
SRP - Inlet	474	404	341	365	296

^a Total N calculated using the sum of TKN and NO₃ + NO₂.

Figure 3. Observed pre and post alum treatment 5 m phosphorus concentrations in Lake Ketchum, WA, Mar 2013–Oct 2016.

Figure 4. Observed pre and post alum treatment 1 m phosphorus concentrations in Lake Ketchum, WA, and observed main inlet phosphorus concentrations from Mar 2013–Oct 2016.

Although^{Table 4}internal loading of P was the largest source of P to the lake, the inlet stream also contributed a substantial load (23% of the total) during 2010–2011 (Table 1), especially during the winter months when flows were high. Inlet TP concentrations from October 2010 through July 2011, when the inflow stopped because of normal summer dry conditions, ranged from 416 to 1060 µg/L.

The 2014 alum treatment was highly effective at reducing internal P loading even though only 66% of the planned dose was applied. Sediment P release was essentially eliminated, decreasing from an average rate of 25 mg/m² per day (2011, 2013, and 2014) to zero (negative rates). Summer hypolimnetic mean TP decreased 87%, from 1427 µg/L in 2013 to 186 µg/L in 2014 (Table 4, Fig. 3), and epilimnetic mean TP concentrations decreased by 88% to 34 µg/L, below the target concentration of 40 µg/L (Table 4, Fig. 4). The TP mass balance model had assumed an 85% reduction in internal loading (SRR), with a predicted epilimnetic mean TP concentration of 46 µg/L for the first year after implementation of the preferred alternative (Table 2). The observed post-treatment epilimnetic mean concentration was similar to the 41 µg/L predicted by the model if internal P loading were eliminated and there was no diffusion from the high hypolimnetic TP levels. The lower than expected result indicated that sediment P release was actually eliminated following the first treatment, which had a direct effect on epilimnetic P concentrations.

TP concentrations were reduced even further in both the epilimnion and hypolimnion following the second alum treatment in March 2015 (Figs. 3 and 4). Summer epilimnetic mean TP decreased an additional 56% from 2014 (to 15 µg/L) for a total reduction of 95% between 2013 and 2015 (Table 4). Internal loading during summer 2015 remained at essentially zero (gross SRR was negative), and summer mean hypolimnetic TP was reduced to 17 µg/L, a 91% reduction from 2014 and a 99% reduction from the 2013 pre-treatment level (Table 4). SRP in the hypolimnion was eliminated, and for most of summer 2015 was at or below the MDL (1.0 µg/L). These reductions in epilimnetic and hypolimnetic TP occurred in 2015 despite winter and spring inlet TP concentrations remaining high (Table 4).

TN also declined markedly following the alum treatments, from mean summer concentrations in 2011 of 1240 µg/L (at 1 m depth) and 4620 µg/L (at 5 m) to concentrations of 660 and 700 µg/L, respectively, in 2015 (Table 4). The TN:TP ratio in the epilimnion increased from 6.9:1 to 44:1. Thus, the alum treatments caused P to be much more limiting, despite a near halving in TN.

The annual water column maintenance treatment on 27 April 2016 reduced both epilimnetic and hypolimnetic TP, which had increased because of the continued high external input of P from the main inlet (Table 4). Mean TP concentrations in the epilimnion and hypolimnion prior to the 2016 treatment were 38 and 34 µg/L (Oct 2015 through Apr 2016), respectively. After treatment, TP averaged 16 µg/L in the epilimnion and 23 µg/L in the hypolimnion (May through Oct 2016). SRP was also reduced following the 2016 treatment, decreasing from a mean of 9.3 and 9.6 µg/L in the epilimnion and hypolimnion to 1.3 and 1.5 µg/L, respectively.

Chlorophyll, transparency, and algal toxins

Summer mean Chl-a in 2011 was 81 µg/L, ranging from 23 to 184 µg/L, with higher concentrations occurring in October after entrainment of P from the hypolimnion. Chl-a in 2013, the year before treatment, was slightly lower, ranging from 2.1 to 166 µg/L with a summer mean of 56 µg/L (Table 4). The highest Chl-a concentrations in 2013 occurred in August and September, earlier in the year than in 2011. Concentrations of microcystin, the liver toxin, ranged from 0.94 to 551 µg/L in 2011 and from 0.54 to 539 µg/L in 2013 (Washington State Toxic Algae Database 2016). These toxin concentrations exceeded the Washington state recreational guideline of 6 µg/L for most of the summer, resulting in public health postings for 24 weeks in 2011 and 19 weeks in 2013.

Following the first alum treatment in 2014, Chl-a concentrations did not improve as dramatically as TP and SRP concentrations. Summer mean Chl-a at 1 m depth was only slightly lower in 2014 than in 2013 (45 vs. 56 µg/L); however, the 2014 mean was strongly biased by one extremely high concentration in June (315 µg/L). Although the high June Chl-a concentration was not accompanied by algal toxins, toxic blooms were present for most of August through November 2014, with microcystin values ranging from 16.1 to 675 µg/L (Washington State Toxic Algae Database 2016).

After the second alum treatment in 2015, Chl-a was greatly reduced, with a summer mean concentration of 17 µg/L (Table 4). This represented a 70% reduction from 2013 and a 62% reduction from 2014 concentrations. Chl-a remained below 5.0 µg/L during summer 2015 until a large bloom, dominated by Aphanizomenon, appeared in September and October, with Chl-a concentrations of 45 and 72 µg/L, respectively. However, there were no algal scums and subsequently no toxic algae advisories for the lake during summer and fall 2015. Following the 2016 annual water column maintenance treatment, Chl-a averaged only 4 µg/L in May through October, a 76% reduction from 2015. There was no large bloom in summer or fall 2016 and no algal scums or toxic algae advisories for the lake.

Transparency prior to the alum treatments was low, with summer (May–Oct) means in the hypereutrophic range, 1.0 m in 2011 and 1.7 m in 2013 (Table 4; Welch and Jacoby 2004). Following the first alum treatment, transparency improved slightly to a summer mean of 2.1 m. After the second treatment, transparency almost doubled (summer mean of 4.0 m). On one occasion in June 2015, transparency reached the bottom of the lake (5.9 m) at the mid-lake station. Transparency in 2016 was similar to that in 2015, with a summer mean of 3.5 m. Maximum transparency in 2016 was 4.6 m, slightly shallower than the maximum reached in 2015.

Discussion

Successive alum treatments in 2014 and 2015 dramatically changed the character of Lake Ketchum from hypereutrophy to low mesotrophy. Between 2013 and 2015, there were decreases of 95% in summer mean epilimnetic TP, 99% in hypolimnetic TP, and 70% in Chl-a, and a 135% improvement in transparency. Toxic algal blooms did not occur in 2015.

Three key points and several additional insights come out of this project. First, the marked water quality changes seem to be primarily the result of the virtual elimination of sediment P release. Prior to the treatments, the SRR averaged 25 mg/m² per day (2011, 2013, and 2014), with maximum rates of 42 mg/m² per day. The observed SRR decreased to zero (negative rate) after the 2014 treatment and remained at zero after the 2015 treatment. Further sediment analysis will be required to determine the proportion of mobile and biogenic P inactivated by the treatments.

Second, the improvements in lake quality occurred despite continued external loading from a legacy agricultural source that flows into the lake averaging >450 µg/L of TP, with 85% soluble as SRP. At least for the immediate post-treatment years, the water quality results indicate that in-lake sediment P inactivation treatments can greatly improve lake quality, even with continued external loading. To ensure these water quality improvements are maintained, annual low-dose water column alum treatments are planned to remove the P that flows into the lake each year from the agricultural source. The first of these maintenance alum treatments occurred in April 2016 and was successful at reducing both epilimnetic and hypolimnetic P, which had increased because of high external loading during October 2015 through April 2016. More years of monitoring data will help determine the longevity of the sediment alum treatments and the benefits of annual water column alum treatments to inactivate continued external loading.

Third, the results also showed the usefulness of a relatively simple 2-layer seasonal mass balance TP model in selecting an appropriate, cost-effective restoration alternative and in predicting the potential effects of restoration measures on the quality of a eutrophic lake with high hypolimnetic P. The calibrated model accurately represented the important P-cycling processes in Lake Ketchum and closely predicted the TP changes resulting from whole-lake alum treatments. Together, the P budget, the model predictions, and observed water quality results (2010–2011) showed that internal P loading was the principal cause for Lake Ketchum's hypereutrophic state, even with the high external loading and the lake's strongly stratified condition. The model also served as an important tool to communicate the costs and water quality benefits of potential alternatives to the lake community and to help secure treatment funding.

Other insights include changes in N concentrations, the evolving impacts to cyanobacteria and Chl-a, and the effects on macrophytes. The 2 alum treatments in Lake Ketchum reduced epilimnetic TN by 47% and hypolimnetic TN by 85%. The decreases observed in TN were most probably the result of reductions in algal biomass that followed from the alum treatments. Reduction in TN was, therefore, an effect of, not a cause of, algal biomass reduction. This conclusion is supported by the change in TN:TP ratio from limitation by N to strong P limitation after the treatments.

The primary goal of the Lake Ketchum alum treatments was to control toxic cyanobacteria blooms. Initial results of the 2014 and 2015 treatments showed positive effects on Chl-a concentrations and algal toxins; however, only 1 year of Chl-a data after each treatment is available, and post-treatment phytoplankton samples have not yet been analyzed. Additional years of monitoring will be needed to evaluate the full effects of alum on algal production and the cyanobacterial community. Also, given the hypereutrophic history of Lake Ketchum, more time may be needed to see a more dramatic reduction in Chl-a concentrations. Cyanobacteria, which have historically dominated the phytoplankton species in the lake, have the ability to store P in reserve before vertical migration through the water column. This cell-bound P would not be available for binding with Al during treatments except for the cyanobacteria cells physically stripped from the water column by the alum floc. After several more annual alum treatments, the readily available P reserve should diminish. Because the external P load is still high, however, episodic blooms, such as the fall 2015 Aphanizomenon bloom, may continue, but hopefully at a smaller scale.

Finally, Lake Ketchum has also shown evidence of switching from a turbid, algal-dominated system to a clear water, plant-dominated waterbody because of the reduced nutrient levels and increases in water clarity. There has already been a strong resurgence of submersed aquatic plant growth. Following the 2014 and 2015 alum treatments, submersed aquatic plants (primarily Potamogeton spp. and Elodea canadensis) were observed in the lake for the first time in several years and have formed dense canopies in places. Submersed aquatic plant growth continued with expanded coverage and density in 2016 as well.

Funding

Funding for the Lake Ketchum restoration was provided by grants from the Washington State Department of Ecology Freshwater Algae Program, by the Surface Water Management Division of Snohomish County Public Works, by a grant from the Snohomish County Stillaguamish Clean Water District, and by property owners at Lake Ketchum.

Acknowledgment

This project would not have been possible without the long-term support of Lake Ketchum residents.