Attempted management of cyanobacteria by Phoslock (lanthanum-modified clay) in Canadian lakes: water quality results and predictions

ABSTRACT

Nürnberg GK. 2017. Attempted management of cyanobacteria by Phoslock (lanthanum-modified clay) in Canadian lakes: water quality results and predictions. Lake Reserv Manage. 33:163–170.

When internal phosphorus (P) loading from the bottom sediments outweighs external P inputs to lakes, lake water quality and cyanobacteria blooms will not respond to external measures alone and therefore require an in-lake restoration treatment. Lake characteristics and governmental regulations do not permit a random choice of such methods. A treatment developed by an Australian research institute (CSIRO) more than 15 years ago has recently been introduced to Canada and is licensed in Ontario. The lake water treatment consists of the application of a phosphate adsorbing and sediment capping agent called Phoslock in North America, a clay (bentonite) that has been systematically amended by the phosphate adsorbing element lanthanum. When applied, Phoslock sinks to the lake bottom where it intercepts the upward flux of internal load from sediment P release. Although the number of monitored Canadian applications is still small, suggestions to optimize restoration effects can be made, including the avoidance of high flushing during application, system isolation, and timing to coincide with high levels of inorganic P in the lake water. Initial results in at least 2 cases (Swan Lake, City of Markham, Greater Toronto Area; and Henderson Lake, Lethbridge, Alberta) were promising despite the unexpected external P input from waterfowl and possibly other external sources. Other applications are planned in a Quebec lake and are being discussed in several other provinces.

KEYWORDS:

• Canadian lakes
• chemical controls < in-lake management
• internal phosphorus load
• lanthanum-modified bentonite (Phoslock)

Most Canadians (∼75%) live in a 161 km band just north of the US–Canada border (travel.nationalgeo-graphic.com/travel/countries/canada-facts/), including the immediate vicinity of the Great Lakes and the St. Lawrence River reach (33%), the central prairies, and the western maritime sections. This region coincides with the favorable growing zones and most fertile soils in Canada. Accordingly, the most enriched Canadian lakes, which are more prone to water quality problems including cyanobacterial blooms, have been previously identified in that region (Chambers et al. 2001). Thus, lake management is especially important in these southern regions for natural (geochemical) and sociological (population density) reasons.

The most obvious sign of lake water deterioration from eutrophication is the increase of potentially toxic cyanobacterial blooms. The number of harmful cyanobacteria blooms (O'Neil et al. 2012) and cyanobacteria dominance within the phytoplankton (Taranu et al. 2015) have been increasing worldwide in north temperate lakes. Ever more studies relate cyanobacterial blooms to increased internal phosphorus (P) loading from lake bottom sediments, also described as “legacy P” from previous external P inputs (Sharpley et al. 2013). In Canadian or US-bordering lakes, relationships were observed between cyanobacteria counts and releasable sediment P fractions (Smith et al. 2011), between cyanobacteria dominance and hypolimnetic total P (TP) concentration influenced by water column stability (Persaud et al. 2015), and between late summer chlorophyll concentrations, mostly from cyanobacteria, and annual internal loading rates (Nürnberg et al. 2013, Nürnberg and LaZerte 2016a). In addition, sediment-derived P aided metalimnetic cyanobacteria to form surface blooms, depending on physical features including internal waves (Pannard et al. 2011) and artificial mixing (Nürnberg et al. 2003).

Suggestions to address internal load management as a potential approach to reduce cyanobacteria blooms (Paerl et al. 2011, Lindim et al. 2015) become particularly important in the face of climate change, which can increase internal loading (Nürnberg et al. 2012) and cyanobacteria blooms (O'Neil et al. 2012, Planas and Paquet 2016). Further, in-lake restoration may be the only management tool available where waste-water treatment and restoration efforts in the catchment basin have minimized external P loading without obtaining acceptable water quality (Lürling et al. 2016).

There are numerous technical and limnological considerations with respect to internal load management (Zamparas and Zacharias 2014), as well as societal and regulatory constraints. For example, the Canadian provinces of Ontario, Quebec, and British Columbia do not favor chemical applications involving aluminum (Al) because of recurring public concern about Al toxicity, especially in regions with poorly buffered waters. Consequently, the P adsorbing and sediment capping agent, Phoslock, may currently be the most promising in-lake treatment for Canadian lakes (Nürnberg and LaZerte 2016b).

Phoslock (lanthanum [La] modified clay) has been applied in about 200 lakes worldwide to treat internal P loading from sediments since 2001 (Copetti et al. 2016). La phosphate mineral phases form so that P retention is increased. For example, P release experiments on intact sediment cores from treated European lakes indicated increased P retention in 8 of 10 lakes and effective control of sediment P release between 2 and 9 years after treatment (Dithmer et al. 2016). A meta-analysis determined a significant median annual TP and soluble reactive P (SRP) concentration decrease in all 15 (TP) and 14 (SRP) examined lakes from a 2-year period before to a 2-year period after a Phoslock treatment, particularly in the fall and winter (Spears et al. 2016).

This note summarizes past Canadian applications and applications planned or proposed in the near future (as of summer 2016). Where possible, lake conditions after the treatment were compared with those before, and TP concentration as an indicator of water quality and trophic state was predicted for a planned treatment. Failure and errors were determined, “lessons learned” are presented, and methods needed to evaluate treatment success and warrant effective applications are discussed.

Data source and methods

Limnological and treatment data were gleaned from the published literature where possible. Because Phoslock has only recently been introduced in Canada, much information was only available in reports and was supplemented by personal communication of the involved parties. Data sources include: Scanlon Creek Reservoir: Moos et al. (2014); Swan Lake: Nürnberg and LaZerte (2016b); Henderson Lake: Yasseri et al. (2014), Nürnberg and LaZerte (2016c), John Derksen of Lethbridge College (pers. comm. 2015), and Staff of the City of Lethbridge (pers. comm. 2015); Lac Bromont: Planas and Vanier (2014) and Dolores Planas, University of Quebec at Montreal (pers. comm. 2016); and Elk Lake: Nürnberg and LaZerte (2016d).

TP analysis was mostly performed with standard methods that included digestion of the water sample and subsequent phosphate analysis by the molybdenum-blue method (Murphy and Riley 1962). In Henderson Lake, however, water samples were analyzed without digestion and results converted to TP by multiplication with a factor of 1.3, considering the lake's trophic state and experience gained in previous studies (Nürnberg 1984). Internal load was determined, and TP was predicted according to methods presented in Nürnberg (2009) and Nürnberg and LaZerte (2016b).

Case studies

To date, 4 systems have been treated with Phoslock in Canada and 1 more is planned for 2017 (Fig. 1, Table 1). All these lakes were treated because of recurrent cyanobacterial blooms and are located within the most populated and fertile southern Canada region, as described earlier.

Figure 1. Location of Canadian lakes with previous or proposed Phoslock as described in Table 1 (scale units are 300 km).

Table 1. Limnological characteristics and Phoslock application information for the Canadian water bodies with previous or planned Phoslock treatment.

Lake characteristics	Scanlon Creek Reservoir	Swan Lake	Henderson Lake	Lac Bromont	Elk Lake
Custodian	Lake Simcoe Region Conservation Authority	City of Markham	City of Lethbridge	City of Bromont	Elk Lake Water Quality Improvements Steering Committee
Province	Ontario	Ontario	Alberta	Quebec	British Columbia
Location (N,W)	44°09′, 79°33′	43°54′, 79°15′	49° 41′, 112° 47′	45°15, 72°40	48°32′ 123°24′
Surface area (m^2)	34,000	54,800	258,000	480,000	1,870,000
Watershed area (km^2)^1	n.a.	0.457	controlled	23.5	8.6
Maximum depth (m)	7.0	4.4	5.5	7.2	18.0
Mean depth (m)	n.a.	1.9	2.6	4.4	9.2
Water residence time (yr)^2	<0.019	∼20, no outflow	1.7	0.25	4.4
Alkalinity (mg/L as CaCO3)	220	118	104	43–56^3	139
Trophic state^4	hypereutrophic	hypereutrophic	hypereutrophic	mesotrophic	oligotrophic
TP Load (g/m^2/yr): external^5		1.9–3.4	0.665	1.752	0.12
TP Load (g/m^2/yr): internal		6.0–10.5	0.652	5.405	0.71–0.94

Note. n.a. = not available.

¹Watershed area is expressed without lake area. Henderson Lake does not have a natural watershed because its inflow is controlled by the St. Mary River Irrigation District.

²Elk Lake water residence time includes downstream Beaver Lake.

³Said Yasseri, Institut Novak, pers. comm. 2016.

⁴Based on growing period trophic state variables, including average TP and chlorophyll concentrations and Secchi disk transparency (Nürnberg 1996).

⁵Lac Bromont external load includes only stream input, not atmospheric deposition and ground water.

The first 2 trial applications were spearheaded by the Lake Simcoe Region Conservation Authority in Ontario and included a stormwater pond (not described here) and a repeat application in an impounded river section (Scanlon Reservoir of West Holland River) in 2008 and 2009. The trial applications in Scanlon Reservoir (Table 1) revealed only immediate positive effects on water quality. Water TP concentration inferred from paleolimnological methods were about 0.1 mg/L in the Scanlon Reservoir without any sustained long-term trend or change with respect to the Phoslock applications (Moos et al. 2014). Under-dosage in 2008 and elevated flushing (Table 1) exasperated by wet summers were proposed causes of the poor treatment performance (Moos et al. 2014).

The treatment has been considered more frequently throughout Canada after the assembling of procedures (Lake Simcoe Region Conservation Authority 2010) and a thorough investigation of potential toxicity (Ontario Ministry of Environment 2009) in the province of Ontario. Further, the lack of proven toxicity in many applications and tests elsewhere (Copetti et al. 2016) has provided confidence in the treatment's environmental safety.

The first full Canadian lake application was conducted in 2013 with the treatment of urban Swan Lake, ON (Table 1), and its effects have been described in detail (Nürnberg and LaZerte 2016b). The significant decrease of average TP from 0.25 mg/L before treatment to 0.06 mg/L in the second post-treatment growing period (Table 2) and related decline in algal biomass indicated treatment success. A lack of response of the algal biomass in the first treatment year was attributed to a relatively late application, when P released from the winter bottom sediments had already been taken up by phytoplankton. Post-treatment TP concentrations were predicted correctly under the assumption of zero internal P loading, supporting the cause of TP decline. An increased population of waterfowl (mostly Canada geese, Branta canadensis), however, suggested an increase of the TP concentration from 0.06 to 0.10 mg/L in future post-treatment years (Table 2). Consequently, efforts to discourage waterfowl and continued water quality monitoring are currently (summer 2016) conducted by the City of Markham to evaluate the need for further treatment.

Table 2. Treatment success, as evaluated by observed and modeled TP concentration before and after treatment, and application details.

Application evaluation and details	Scanlon Creek Reservoir	Swan Lake	Henderson Lake	Lac Bromont	Elk Lake
TP concentration (mg/L), mixed layer, growing period
Observed: before	0.02–0.07	0.25	0.22	0.02	0.009
Observed: after	no trend	0.06	0.024^1	n.a.	n.a.
Predicted: before	n.a.	0.27	0.20	n.a.	n.a.
Predicted: after (longterm average)	n.a.	0.10^2	0.10	n.a.	n.a.
Application details
Dates	2 times: 2008; once: 2009	29 April–1 May 2013	25–29 Apr 2016	Expected 2017	In discussion
Total Phoslock applied (t)	3.2; 15.0	25.2	64	174	tbd
Areal application rate (kg/m^2)^3	0.10; 0.44	0.46	0.25	0.36	tbd
	River section	Increasing waterfowl #	Many different treatments: herbicides, mixing, aeration, grass carp	Summer and fall cyanobacterial blooms	Summer and winter cyanobacterial blooms

Note. n.a. = not available; tbd = to be determined.

¹Average for 10 dates and 5 sites between 3 May and 14 Sep 2016.

²Includes P input from increased water fowl abundance.

³The areal application rate is computed as total applied Phoslock mass divided by lake surface area.

Preliminary results at the time of the final assembling of this note indicate that despite the apparently successful management of resident geese in the summer 2016, TP concentrations were as high as and Secchi transparency as low as those measured in the pre-treatment summer, but no surface blooms were observed. Resident geese were abundant until management in 2016, and large flocks of migratory geese still frequent the lake. Present and historic site inspection led to further possible P sources provided by a former landfill and a waste disposal site that could contaminate groundwater and runoff. Such potential effects must be considered in context with weather patterns, especially the enhanced precipitation in 2016.

The most recent Phoslock treatment was completed in April 2016 in the urban, highly eutrophic Henderson Lake, Alberta (Tables 1 and 2). Henderson Lake is a modified depression in the prairie landscape surrounded by parklands, a golf course, and a botanical garden; its inflow is controlled by a water agency (St. Mary River Irrigation District). A variety of techniques have been applied, including the aquatic herbicides Reglone and Reward, Liquid Live Micro Organisms, annual additions of grass carp (Ctenopharyngodon idella), 5 Solar Bees (solar powered water column mixing devices), and a diffuse aeration system to control rooted plants or reduce algae. None of these treatments has reduced TP concentration, however, and cyanobacteria continued to negatively affect lake water quality.

Phoslock was applied at a relatively low rate in Henderson Lake (Table 2) after the application rate was determined by sediment extraction of the mobile P fraction similar to that conducted in other studies (Meis et al. 2012, Bishop et al. 2014). A preliminary long-term mass balance analysis similar to that used in Swan Lake (Nürnberg and LaZerte 2016b) predicted a 50% decrease in TP concentration as a maximum Phoslock treatment effect (zero internal load; Table 2) because of the equal size of the estimated internal and external P loads (Table 1). The observed TP decreased much more, however, down to ∼12% of the pre-treatment concentration in the first post-treatment growing period. The average TP for 10 dates and 5 sites in 3 May 3 to 14 September 2016 was 0.024 mg/L in the mixed surface water (Table 2) and similar at 0.028 mg/L in the bottom water, indicating zero internal load.

The next Canadian Phoslock treatment is scheduled for Lac Bromont, Quebec, for 2017. Lac Bromont has been studied for more than 5 years (Planas and Vanier 2014), and mass balances estimated that internal load as P accumulated in the fall hypolimnion is 3 times that of the annual external load (Table 1). In this relatively shallow but stratified dimictic lake, the cyanobacteria consisted mainly of metalimnetic Planktothrix species (rubescense and agardhii) controlled by internal nutrient load and light availability influenced by internal waves (Pannard et al. 2011). Because there was a strong correlation between annual cyanobacterial biomass and hypolimnetic TP mass, it is likely that internal load drives the recurring cyanobacterial blooms despite the comparably low epilimnetic TP concentration of 0.02 mg/L (Planas and Vanier 2014). An intermediate application rate has been proposed (Table 2).

A potential Phoslock treatment has been discussed to abate summer and winter blooms of toxic cyanobacteria in monomictic Elk Lake, British Columbia (Nürnberg and LaZerte 2016d). The lake is an important and valuable asset to the British Columbia capital region of the Greater Victoria area with >1.5 million visitors per year (2014 estimate) and has traditionally hosted “polar bear swim” events in January. Unacceptable potential toxicity from cyanobacteria, as well as concern for lake users including rowers, swimmers, pets, and other stake holders throughout the year, has sparked discussions about possible restoration techniques. The evaluation of techniques to address the internal P loading was deemed especially important because surrounding lakes have experienced similar deterioration.

Although Elk Lake is oligotrophic during the growing period and the current external load is small (Table 1), TP, SRP, and chlorophyll concentrations drastically increase after thermocline erosion and fall mixing (Fig. 2). The recent internal load estimated from hypolimnetic increases throughout the growing period was 6–8 times larger than the external load, probably caused by accumulation of legacy P in the sediment from intense agricultural and anthropogenic P inputs in the past. A Phoslock application, perhaps specific to the sediment overlain by anoxic water (below 7 m) if funding resources are limited, would curtail sediment release and fertilization of the cyanobacteria blooms after destratification at the end of the year. This treatment method is expected to also prevent early winter and spring blooms and decrease phytoplankton biomass year-round.

Figure 2. TP, SRP, and chlorophyll in surface water at Elk Lake, main station (Feb 2014–Aug 2015).

Dosage, costs, example Elk Lake

Phoslock contains 5% La by weight, and P is adsorbed by La at a molar ratio of 1:1 (Ross et al. 2008). Dosage is calculated from the amount of Phoslock necessary to inactivate the potentially available P in the sediment and in the water column (Meis et al. 2012, Bishop et al. 2014). The releasable P in the sediment (mobile P) is quantitatively most important and estimated by sediment fractionation (sum of pore water P, iron-bound P, and labile organic P; Reitzel et al. 2005).

The provided material cost of Phoslock was C$3100/t (metric ton) in 2015, and the application cost was C$200/t (Phoslock Water Solutions Ltd., 22 Sep 2015, preliminary cost estimate for a potential treatment of Elk Lake), but the price of Phoslock is expected to drop in the future as production volumes increase. A mid-range dosing rate of 4.6 t/ha or 0.46 kg/m² was initially suggested at a cost of C$15,216/ha (rates ranged from 1.4 to 6.8 t/ha in 18 lake treatments; Spears et al. 2016). At this rate, the treatment of an assumed active area below 9 m would total C$1,686,000, and a treatment below 7 m would cost C$1,925,000 (Nürnberg and LaZerte 2016b). These costs include a small dose (0.3% of total dose) to treat P in the water column.

To warrant sufficient material in an effective application, but also to avoid expensive over-dosing, a more exact determination of the minimum effective dose is preferable and more economically feasible. Accordingly, the Elk Lake Water Quality Improvements Steering Committee commissioned sediment sampling and analysis. The mobile P average concentration of cores taken at 7 locations within the anoxic hypolimnion at 9–16 m depth (Yasseri and Van Goethem 2016) suggest that the initially proposed dosage can bind all mobile P in a sediment layer of 0–7.0 cm depth. This dosage should therefore create a sufficient barrier to sediment P release in the target area.

Discussion

The information on the described treatment projects emphasizes the importance of knowing the size of internal compared to external load (Nürnberg 2009). Although the study lakes span trophic states from hypereutrophy to oligotrophy, they all experience cyanobacterial blooms. Because trophic state including lake TP concentrations alone does not seem to control these blooms, the contribution from specific P sources and specific P forms of different bioavailability must also be considered. Similarly, much of the recent deterioration in Lake Erie hypoxia and cyanobacterial blooms has been attributed to increased SRP (but not TP) load (Scavia et al. 2014). The most bioavailable sources include fertilizer and treated or untreated discharges from waste water treatment plants, septic systems, or feedlots as external load, and the release of legacy P from enriched sediments as internal load. Because internal load is released as phosphate directly into the lake water without further dilution, its bioavailability is almost 100% (Nürnberg and Peters 1984). When highly bioavailable external sources have been controlled, as is increasingly occurring in western developed countries, the abatement of internal loading promises a decrease in cyanobacteria blooms.

Without the abatement of large external P inputs, even the most direct restoration technique of sediment dredging may prove ineffectual. Settling suspended particles from river inflows caused P release in sediment cores of Lake Chaohu, even after experimentally simulated dredging (Liu et al. 2016). In some cases, external load cannot be realistically managed. For example, internal load abatement was found to be the most feasible and cost-effective solution for long-term management of water quality in the Minneapolis Chain of urban lakes in Minnesota (Huser et al. 2016b). In other cases, the external loading occurs directly into the lake (e.g., bathers or waterfowl), and a Phoslock treatment can bind the SRP in the water as well as the newly accumulated sediment P (Epe et al. 2016, Nürnberg and LaZerte 2016b).

It also has to be considered that internal loading may be enhanced by bioturbation, especially by bottom dwelling fish such as carp (Cyprinus carpio). Although mobile P in the sediment surface layer can be mechanically disturbed by fish (Huser et al. 2016a), it would still be bound by any added Phoslock. Phoslock effectively reduced the P fluxes across the sediment–water interface during controlled resuspension in microcosm experiments and so can be expected to also work well in systems where sediment is resuspended by wind disturbance or bottom dwelling fish (Yin et al. 2016).

While all the study lakes are within the most populated zone around the US–Canadian border, lakes in the remaining and largest part of Canada may also have to be considered because cyanotoxins have been increasingly observed throughout Canadian lakes (Pick 2016). Most of this area is comprised of Precambrian shield, mountains, or the lowlands and tundra (to the north). In these regions, lakes are typically oligotrophic with low productivity and high transparency, or tea-stained by humic and fulvic acids when wetland-fed, but even some oligotrophic to mesotrophic lakes have recently displayed increased cyanobacterial biomass and blooms (Winter et al. 2011). The investigation of nutrient sources, including the separation between natural, anthropogenic, and internal loading, is needed before the suitability of in-lake remediation in such relatively remote areas can be assessed.

In general, any restoration endeavors are constrained by political and societal matters in addition to the technical aspects. Lake management that includes in-lake restoration techniques may be necessary to control future cyanobacteria-related water quality problems. To ensure positive results, site-specific studies and lake management plans may be necessary (Zamparas and Zacharias 2014).

Conclusions

•	For a successful application of Phoslock and other in-lake treatment, external P load, especially of bioavailable SRP, should be much smaller than internal load to avoid further increase of mobile sediment P.
•	Therefore, external load evaluation and control should consider P bioavailability and less obvious P sources (e.g., waterfowl management in Swan Lake, groundwater).
•	A chemical treatment of fast flushed systems is not promising (e.g., Scanlon Reservoir, trial application) because external input continues, and the treatment material may be flushed out before settling.
•	Low dosage is not effective (Scanlon Reservoir 2008) because trial dosing may not sufficiently intercept P release.
•	Control of application timing to coincide with a high bioavailable P level in the water column ensures immediate and maximal responses (e.g., Swan Lake, first summer).

Acknowledgments

Data, unpublished reports, and discussions were generously provided by Prof. Emeritus, Dolores Planas, University of Quebec at Montreal; Anne Joncas of the Bromont Lake Association; John Derksen of Lethbridge College; Staff of the Cities of Lethbridge and Markham; Said Yasseri, Institut Dr. Nowak, Germany; and Nigel Traill of Phoslock Water Solutions Ltd. The numerous helpful comments by Bruce LaZerte, Ken Wagner, and other reviewers are appreciated.

Funding

Funding for data analysis and reports by the author on specific lakes was provided by the City of Markham, the Elk Lake Water Quality Improvements Steering Committee, and Phoslock Water Solutions Ltd.