Evaluation of preventative algaecide treatments for cyanobacterial resting cells in sediments of a central US lake

Abstract

Kinley-Baird CM, Smith EF, Calomeni AJ, McQueen AD, Gusler GO, Boyer M, Decker KN, Clyde GA. 2023. Evaluation of preventative algaecide treatments for cyanobacterial resting cells in sediments of a central US lake. Lake Reserv Manage. 39:340–355.

Cyanobacteria can overwinter in sediments of aquatic systems as resting cells and contribute to harmful algal bloom (HAB) resurgences. We evaluated field-scale preventative algaecide treatments in April 2022 in terms of effectiveness for decreasing viability of cyanobacterial resting cells, with the goal of minimizing planktonic growth in season. Two consecutive treatments of a granular peroxide-based algaecide (GreenClean PRO) were applied (48 h apart) in a Kansas, United States, lake. Treatments were applied to a 31.5 ha treatment zone, separated from a 12.9 ha control zone using a silt curtain. Treatment design was informed by laboratory algaecide efficacy experiments using site-collected samples. In-lake treatment performance was evaluated based on comparisons of planktonic cyanobacterial densities in the treatment and control zones measured every 2 weeks for 4 months, then monthly for 2 additional months. Other lines of evidence included resting cell densities and recruitment viability in sediments 3 d after treatments were performed. For 9 of 11 sampling events from May to October, average planktonic cyanobacterial densities trended lower in the treatment zone than in the control zone (ranging from 6 to 97% lower). Declines in sediment cell density and recruitment of resting cells in laboratory assessments were not measurable 3 d after treatments were performed. Additional studies will be necessary to strengthen the database supporting use of this management strategy. However, there may be an opportunity to expand the model for how algaecides are used to manage HABs more effectively and efficiently in freshwater resources.

Keywords:

• Akinetes
• chemical treatment
• field demonstration
• harmful algal blooms
• peroxide

Harmful algal blooms (HABs) composed of toxin-producing cyanobacteria pose severe threats to the uses and economic values of freshwater resources throughout the United States (Brooks et al. 2016). Given the increased frequency and severity with which HABs are occurring (Hallegraeff 1993, Paerl and Huisman 2009), both short-term and long-term management approaches are necessary to minimize health risks and maintain uses and operations in impacted systems. The use of US Environmental Protection Agency (USEPA)-registered algaecides is a short-term management approach to decrease cell densities of HABs and associated toxin levels either after site monitoring indicates evidence of growth or following visual observations of cells accumulating at the water surface. Proactive strategies using algaecides have been applied with success (Huddleston et al. 2015, Calomeni et al. 2017), and are often utilized in lake management practices, with efforts primarily focused on using strategic monitoring programs to trigger treatments early in spring and summer growth cycles. There may also be an opportunity to broaden algaecide use as a preventative tactic, given what is known about the ability of specialized resting cells of pelagic cyanobacteria to survive and accumulate in sediments, and evidence that these cells contribute to HAB formation (Calomeni et al. 2022, 2023a).

Genera of cyanobacteria including those of the order Nostocales (e.g., Aphanizomenon, Dolichospermum, and Raphidiopsis) and order Chroococcales (e.g., Microcystis) can remain quiescent in sediments during colder seasons and then “germinate” and reproduce when conditions are again adequate for growth (Fay 1988, Adams and Duggan 1999, Kaplan-Levy et al. 2010). Collectively, cyanobacterial akinetes (found in Nostocales) and vegetative cells that remain dormant in sediments are referred to here as “resting cells.” Prior studies have shown that recruitment (i.e., transfer from sediments to water) of these cells in warmer seasons can be a significant driver for planktonic growth and formation of HABs (Kaplan-Levy et al. 2010, Cirés et al. 2013, Kitchens et al. 2018). Thus, “seed beds” of resting cells can be considered a target assemblage for management, and such tactics would be aimed at minimizing viability of these cells in sediments prior to planktonic recruitment each year.

Our overarching hypothesis was that this management approach could result in delayed onset and/or decreased severity of HABs, as demonstrated by a decrease in planktonic cell densities of cyanobacteria during months when HABs are anticipated to occur. However, preventative algaecide treatment of cyanobacterial resting cells has not been studied beyond the bench scale, and potential outcomes of this approach are currently unknown. Since benthic algaecide application technology is already available (e.g., weighted drop hoses and granular algaecide formulations), a field-scale evaluation of preventative algaecide treatments was readily achievable. We identified an appropriate study site by confirming presence and viability of resting cells in sediments (Calomeni et al. 2022, 2023a) of a historically HAB-impacted site and monitoring those parameters for several months before treatment. Bench-scale algaecide efficacy experiments were performed using site-collected samples to identify an effective algaecide formulation, concentration, and frequency, providing appropriate replication to discern differences between candidate treatments before implementing at full scale (Calomeni et al. 2015, Geer et al. 2017, Kinley-Baird et al. 2021). Following identification of effective treatment parameters, a performance monitoring plan was designed to measure effectiveness from the preventative treatments.

Our overall objective was to evaluate the performance of 2 consecutive hydrogen peroxide-based algaecide (GreenClean PRO) treatments (48 h apart) targeting cyanobacterial resting cells in sediments of Milford Gathering Pond (Junction City, KS). Specifically, our objectives were to (1) measure responses of resting cells to algaecide treatments within 1–3 d after treatment in terms of sediment cell densities and recruitment potential (via laboratory incubation studies), and (2) measure and compare cell densities of target planktonic cyanobacteria in the treated and untreated areas of the lake over time from May through October 2022.

Study site

Milford Gathering Pond, referred to here as “the lake,” has an area of 44.5 ha and is adjacent to Milford Reservoir in Junction City, Kansas (Fig. 1), with an average depth of 1.6 m and a maximum depth of 3.8 m. The lake is managed by the US Army Corps of Engineers (USACE) Kansas City District (NWK), in terms of water uses and budget, and is used by the public for recreation (swimming beach) and fishing. The Kansas Department of Wildlife and Parks (KDWP) leases land on the northern side of the lake to operate a fish hatchery (Fig. 1). The hatchery uses water from the lake as well as from groundwater wells for operations and approximately 95% or more of the water used in the hatchery gets discharged to the lake. The lake also receives inflows from weep wells extending from the dam of Milford Reservoir that carry dam seepage water. Inflows from hatchery effluent and weep wells are in the western basin, while the outflow is located at the eastern shore of the eastern basin. Inflows from weep wells are variable and based on the water level of Milford Lake, but are generally 0.013–0.019 m³/sec. Inflows from the fish hatchery vary based on operational water usage but are on average 0.038 m³/sec originating from groundwater wells and 0.108 m³/sec originating from the lake (between May and Oct only). All inflows have similar pH and conductivity, and all contain relatively high levels of nitrogen and phosphorus (data in supplement). The outfall of the lake discharges into a small stream that flows into the adjacent Republican River. Prevailing winds are from the south/southwest. Average water temperatures measured in the lake in the fall of 2021 ranged from 6.75 to 18.4 C, in the winter of 2022 ranged from 8.1 to 9.1 C, in the spring of 2022 ranged from 14.7 to 28.1 C, and in the summer of 2022 ranged from 23.6 to 29.4 C.

Figure 1. Map showing treatment and control zones, sampling sites, and silt curtain within Milford Gathering Pond. Further detailed explanation of sampling sites is included in the supplement.

The lake has historically experienced HABs in warmer months, with the highest measured microcystin concentration since 2019 being 400 µg/L (KDHE 2022). Total microcystin concentrations often exceed the USEPA recommended recreational ambient water quality criterion of 8 µg/L (USEPA 2019, KDHE 2022). HAB occurrences have resulted in frequent swimming beach closures, particularly in 2019, when the beach was closed for several consecutive months in the summer. Due to this lake experiencing HABs annually, we anticipated a high probability of resting cells being present in site sediments. Historical data available from 2019 through 2022 show dominant HAB genera in the lake have included Aphanizomenon, Dolichospermum, Microcystis, Planktothrix, and Raphidiopsis. HABs are typically reported beginning in June and can persist through late October (KDHE 2022).

Milford Gathering Pond has a narrow channel that separates the lake into 2 basins. Given the lake’s separate basins, this presented an opportunity to divide the lake and assess peroxide-based algaecide treatment in one basin while leaving the second basin untreated as a reference (identified throughout this article as treated and untreated zones; Fig. 1).

Materials and methods

Timeline of tasks in study

This study involved multiple steps to accomplish the ultimate objective of evaluating field-scale algaecide treatments targeted at resting cyanobacteria cells in sediments of Milford Gathering Pond. First, site monitoring was performed between September 2021 and March 2022 to confirm the presence and viability of resting cells, and to measure densities in sediments over time. Then, laboratory-scale experiments were conducted with site-collected samples in February 2022 to measure and compare effectiveness of several peroxide-based algaecides for control of cyanobacterial resting cells. In these experiments, resting cells in sediments were isolated and exposed to several algaecides at a range of treatment concentrations using filtered site water to discern extent of recruitment following treatment as compared to an untreated control. The goal of these experiments was to identify an effective formulation, concentration, and frequency of treatment for use at the field scale. In-lake algaecide treatments were performed in April 2022 with monitoring occurring immediately prior and several days after to evaluate performance of the treatments in the short-term. Finally, surface water monitoring was carried out between May and October 2022 to measure and compare planktonic cell densities in the treated and untreated zones (i.e., basins) of the lake as another metric for treatment performance.

Site monitoring: September 2021–March 2022

Site monitoring was initiated in September 2021, 7 months before the anticipated time of treatment. The goals for the strategic monitoring were (1) to confirm the presence of cyanobacterial resting cells in sediments, (2) to enumerate and evaluate viability of the resting cells (i.e., ability to germinate and transfer into the water column under ideal growth conditions in the laboratory), and (3) to monitor trends in resting cell density and viability over the winter to evaluate for potential natural attrition. If resting cells were present, persisted through the winter, and had measurable recruitment potential, these lines of evidence would support the potential contribution of resting cells to HAB formation at this site and confirm the lake as an appropriate candidate for preventative treatment evaluations. Monitoring events took place in September 2021, November 2021, February 2022, and March 2022.

At the first sampling event (September 2021), sediment and water samples were collected at 4 sites along one defined transect in each of the planned treatment and control zones. At each of the 4 sampling sites, one sediment sample was collected using a petite ponar dredge, and water near the sediment–water interface (for use in incubation studies) was collected using a Van Dorn bottle. Thus, there were 4 sediment samples and 4 bottom water samples collected in each zone. One surface water sample was collected in each zone (planned treatment and planned control) for identification and enumeration of algae and cyanobacteria present in the water column. In situ measurements of Secchi disk transparency, temperature, pH, and dissolved oxygen accompanied each water column sample (data in supplement). Based on results from this sampling event, relatively high variability in resting cell densities among sites within both the planned treatment and planned control zones was observed. The sampling protocol was redesigned to increase the spatial area sampled in each zone and to increase the area represented within each sample.

For the 5 remaining sampling events in which sediments were collected, 3 sites were identified in each of the planned treatment and planned control zones. At each of the 3 sites, 3 separate sediment samples were collected using a petite ponar dredge, with the upper 2 cm of each collected using a spatula and composited into one sample for analysis. Water samples were collected as previously described, with the exception that surface water samples were then collected at every site. Thus, in the revised sampling protocol, 3 sediment samples (each sample being a composite of 3 grab samples per site), 3 bottom water samples, and 3 surface water samples were collected from each zone. From each surface water sample, part was aliquoted into a sample bottle for taxonomic analysis, and the remainder was aliquoted into sample containers for analysis of cyanotoxins and selected nutrients according to methods outlined in KDHE (2020, data in supplement). Once samples were collected, they were stored in sealed containers and placed in coolers with ice either for overnight shipment to the Aquatic Control laboratory in Seymour, Indiana (taxonomy samples), or for direct delivery to Kansas Department of Health and Environment (KDHE) laboratories (for water chemistry and toxin analysis). Two HOBO 64k temperature and light data loggers (Onset HOBO, Bourne, MA) were placed 0.3 m below the water surface in the littoral area in the planned treatment (39°4′56.66ʺN, 96°53′5.14ʺW) and planned control (39°4′37.96ʺN, 96°52′40.00ʺW) zones of the lake to collect data throughout the treatment and post-treatment monitoring (results in supplement).

For taxonomic analysis and recruitment testing, sediment samples were prepared and analyzed for resting cell enumeration according to Calomeni et al. (2023a), with the exception that a gridded Sedgewick Rafter counting chamber (Wildco; Yulee, FL) was used for enumeration as opposed to a Palmer–Maloney chamber. Cells were enumerated in 15 of 50 columns on the chamber grid using a Motic Panthera C2 (San Antonio, TX) compound microscope at 200× magnification. The same columns were used consistently for all samples. A correction factor was applied to final cell counts based on the fraction of the total volume of the counting chamber that was enumerated, allowing units of measurement to be expressed in terms of cells per milliliter. Surface water samples were analyzed using the same counting chamber and microscope. Incubation studies were performed using methods and parameters from Calomeni et al. (2023a). All incubation studies to assess viability of resting cell assemblages used filtered site water (using 0.45 µm nitrocellulose filter paper) to prevent confounding results from existing planktonic assemblages on final cell density measurements.

Laboratory algaecide efficacy experiments: February 2022

Samples were collected for the laboratory algaecide efficacy experiment during the routine sampling conducted on 16 February 2022, using procedures described in the prior section. In addition to the samples regularly collected, 20 L of site water was collected from the surface water for use as overlying water in algaecide exposure chambers. After routine analyses were completed, all sediment samples from the planned treatment zone were composited and homogenized by manual mixing for use in the experiment.

Bench-scale algaecide efficacy experiments were initiated within 1 week of sample collection from the lake. Four hydrogen peroxide-based algaecides were evaluated. Three formulations were granular and contained an active ingredient of sodium carbonate peroxyhydrate (SCP; GreenClean PRO, PAK 27, and Phycomycin SCP), and one formulation was liquid with active ingredients of hydrogen peroxide and peroxyacetic acid (GreenClean Liquid 5.0). Due to the amount of time necessary to measure cell densities at the end of these experiments, algaecide treatments were initiated across 2 consecutive days. Treatments for 2 algaecide formulations (PAK 27 and GreenClean PRO) were initiated on day 1 and the remaining 2 (Phycomycin SCP and GreenClean Liquid 5.0) were initiated on day 2. Untreated controls were prepared for each day. For each algaecide, 4 exposure concentrations (n = 3) within allowable label range were evaluated (Table 1). An additional set of treatments had algaecides applied twice, 48 h apart. Treatments applied twice involved the 2 higher exposure concentrations of each algaecide (Table 1). The purpose of evaluating 2 consecutive treatments was to discern any added effectiveness. Label directions for these hydrogen peroxide-based algaecides allow for applications that are 48 h apart regardless of surface area treated (except for sites containing dense blooms).

Table 1. Treatment design for laboratory algaecide efficacy experiments.

Treatment	Exposure concentrations (n = 3)
Negative control	n/a—quartz sand and filtered site water
Untreated control	n/a—site-collected sediment and filtered site water
Phycomycin SCP	3, 5, 7*, 10* mg H2O2/L
PAK 27	3, 5, 7*, 10* mg H2O2/L
GreenClean Pro	3, 5, 7*, 10* mg H2O2/L
GreenClean Liquid 5.0	4, 10, 15*, 22* mg H2O2/L

n/a - not applicable

^* Two treatment frequencies were used for these concentrations in separate exposure chambers. Exposure chambers receiving one treatment received the exposure at time 0 h and exposure chambers receiving 2 treatments received the exposures at time 0 h and 48 h.

Treatments were prepared according to Calomeni et al. (2023b). Briefly, 10 g of homogenized wet sediment was weighed and carefully added to a 250 mL borosilicate beaker. Then, 150 mL of filtered site water (using 0.45 µm nitrocellulose filter) was carefully added over the sediment layer. Negative controls (n = 3) were prepared with 10 g of quartz sand and filtered site water, to test whether filtration was effective at removing cyanobacteria cells from the site water used in experiments. Untreated controls were prepared in the same way as treatments, but without algaecides being applied. Beakers were placed in a dark refrigerator set at 13 C for 24 h to allow sediments to settle. The purpose for refrigeration was to target a water temperature similar to that anticipated in the lake around the time of algaecide treatments. Actual water temperatures measured in the lake on the day before treatment ranged from an average of 13.2 C near the sediment–water interface to 13.3 C at the surface.

Algaecide treatments were initiated after the 24 h settling period. Stock solutions of algaecides were prepared and amended in appropriate amounts to achieve targeted exposure concentrations (Table 1). For the granular peroxide-based algaecides, stock solutions were prepared by dissolving granules in distilled water using a magnetic stir plate and stir bar. Stock solutions were prepared rather than applying the algaecides in granular form because the scale of the beakers was too small to accurately weigh the granular mass needed. For the liquid algaecide, GreenClean Liquid 5.0, a stock solution was also prepared using distilled water.

Immediately after treatments were applied, samples were collected and stabilized for hydrogen peroxide analysis to confirm exposure concentrations (Klassen et al. 1994, Kinley et al. 2015). All treatment and control beakers were then transferred to a refrigerator set to 13 C and light intensity of approximately 100 LUX (to simulate field conditions at the sediment–water interface anticipated at the time of treatment). After a period of 48 h, beakers receiving one algaecide exposure were moved to an incubator (23 ± 2 C, 16 h light:8 h dark, 2600 lux) to allow time for responses to manifest and any remaining viable cells to germinate and grow (i.e., incubation studies). Remaining beakers were treated a second time (with exposure concentrations confirmed again) and placed back in refrigeration for an additional 48 h before moving to incubation. All beakers were placed in incubation for 18 d (after the 48 h exposure period) to provide conditions that would promote germination and growth if resting cells were viable. After incubation, cyanobacterial cell densities were measured in overlying site water using a Neubauer improved hemacytometer and microscope as previously described.

Curtain installation: April 2022

Several days before the in-lake treatment, a heavy-duty turbidity curtain (type II, 6700 g/m² coated PVC; Texas Boom Company, Houston, TX) was installed and anchored (using dual Duckbill 88-DB1 Earth Anchors on both sides of curtain) to one shore of the natural 25 m pinch point between the control and treatment zones (39°4′41.00ʺN; 96°52′54.57ʺW). It was deployed and anchored to the opposite shore immediately before the first algaecide application. Routine checks of the curtain were performed during the field sampling events occurring every 2 weeks. The curtain extended from above the water surface to the sediment floor. No apparent significant breaks or tears were observed during the deployment, so water flow between basins was anticipated to be minimal.

In situ algaecide treatments: April 2022

Details about how the algaecide treatment parameters (e.g., concentration and frequency) were selected are described in the Results section. Algaecide applications of GreenClean PRO were performed on 19 April 2022 and 21 April 2022 at Milford Gathering Pond using a 5.5 m Carolina Skiff fiberglass boat (Model 2180) and a 4.9 m Alumitech airboat, each with 2 custom-built granular spreaders located at the bow that broadcast spread algaecide granules across the water surface. The spreaders were calibrated prior to treatment to dispense product at targeted rates in tandem with defined boat speed and measured swath width such that 224 kg/ha was applied to the treatment zone for the first treatment (112 kg per 0.3 m of depth for the bottom 0.6 m of the water column). Boat speed was then adjusted for the second treatment to apply 173 kg/ha (86.5 kg per 0.3 m of depth for the bottom 0.6 m of the water column). Treatment transects were created 3 m apart (i.e., 3 m swath width) using ExpertGPS Pro (v 8.23) and uploaded onto onboard Lowrance units for drivers to follow. The Carolina Skiff boat was used to treat deeper areas (i.e., >1.2 m), while the airboat was used to treat shoreline and shallow areas. On the first day of treatment, 7076 kg of algaecide was applied to the 31.5 ha treatment zone. On the second day of treatment (48 h after), 5443 kg of algaecide was applied to the same area.

Performance monitoring: April–October 2022

Pretreatment samples were collected on 18 April 2022, 1 d before the first in-lake treatment. The first post-treatment monitoring event was conducted on 22 April 2022 (3 d after the first treatment and 1 d after the second treatment). At each sample collection, water and sediment samples were collected from near the same locations as in earlier sampling events in both treatment and control zones. The objective for these sample collections was to evaluate effects to resting cell densities (in sediments) and recruitment potential following the treatments. Surface water grab samples were also collected from 4 newly established sites (3 littoral and one open water) in each zone, for identification and enumeration of cyanobacteria, as these would serve as sites where samples were collected for longer term monitoring in the water column every 2 weeks from May through August, with 2 additional monthly sampling events in September and October. Samples were shipped, processed, and analyzed as previously described for routine monitoring events.

Statistical analyses

A Shapiro–Wilk test was first performed to test for normal distribution of the data for laboratory algaecide efficacy experiments (α = 0.05). Based on these results, data were normally distributed and thus a one-way analysis of variance (ANOVA) was performed for comparisons among all algaecides for the single-dose treatments. Homogeneity of variance was tested for using Brown–Forsythe tests and standard deviations were not significantly different (P = 0.092; α = 0.05). For each algaecide individually, Dunnett’s multiple comparisons tests were used to compare each treatment to the untreated control (α = 0.05). The goal for this analysis was to discern treatment concentrations resulting in significant differences from the untreated controls. Since the lowest cell densities were measured following treatments of the highest concentrations of the granular algaecides, the highest treatment concentrations from each algaecide in single-dose experiments were compared using an ANOVA and Tukey’s multiple comparisons test (α = 0.05), to discern significant differences among cyanobacterial responses when algaecides were applied at maximum label concentration. Finally, single-dose and double-dose treatments were compared among the same concentrations for each algaecide using an ANOVA and Tukey’s tests to evaluate differences in responses based on treatment frequency. Data from Phycomycin SCP double-dose treatments did not meet normality criteria. Therefore, a Kruskal–Wallis ANOVA was used followed by Dunn’s multiple comparisons tests to compare single-dose and double-dose treatments for Phycomycin SCP. All statistical analyses were performed using GraphPad Prism version 9.4.1 for Windows (GraphPad Software, San Diego, CA).

Results

Pretreatment site monitoring

Analysis of samples collected in 2021 (Sep and Nov) and 2022 (Feb and Mar), prior to April 2022 treatments, confirmed that cyanobacterial resting cells were present and viable in site sediments of Milford Gathering Pond and had the potential to contribute to HAB formation (Fig. 2A). In September 2021, average cell densities of resting cyanobacteria in sediments were 77,800 cells/g (wet sediment) in the planned treatment zone and 58,700 cells/g in the planned control zone (Fig. 2A). Laboratory incubation studies to assess viability of these assemblages resulted in averages of 38,400 and 14,700 cells/mL transferred from sediments to water for planned treatment and control zones, respectively (Fig. 2A). Resting cell densities were highest in samples collected in February 2022 (Fig. 2A), with averages of 327,100 and 275,100 cells/g for planned treatment and control zones, respectively. Planktonic cell densities following incubation studies for these samples were 157,400 cells/mL (planned treatment zone) and 13,900 cells/mL (planned control zone; Fig. 2A).

Figure 2. Pretreatment evaluation of average resting cell densities (n = 3) in site-collected sediments and average planktonic cell densities (n = 3) in overlying water following 14 d laboratory incubation studies (A) and average cell densities (n = 3) of planktonic cyanobacteria in situ at time of sampling (B) at Milford Gathering Pond. Error bars indicate ±1 SD. Samples were collected from sites W1, W2, and W3 for treatment zone and E1, E2, and E3 (shown in Fig. 1) for control zone.

In surface water samples collected from the lake, planktonic cell densities of cyanobacteria peaked in November, ranging from approximately 3,000,600 to 3,639,200 cells/mL in planned treatment and control zones, respectively (Fig. 2B). Average planktonic cell densities decreased in February and March, with lowest measurements occurring in March and ranging from 37,100 cells/mL in the planned treatment zone to 111,300 cells/mL in the planned control zone (Fig. 2B). Prominent genera detected in these sampling events included Aphanizomenon, Dolichospermum, and Microcystis.

Laboratory algaecide efficacy experiments

Due to the high variability among replicates in single-dose treatments, there were no significant differences in planktonic cell densities (following 18 d incubation) between treatment concentrations and untreated controls for any of the algaecides evaluated (P = 0.17 to 0.85; Fig. 3). However, the highest concentration evaluated for each granular algaecide (10 mg/L as H₂O₂; maximum allowable per label) resulted in the lowest average planktonic cell densities, which ranged from 8750 to 50,000 cells/mL. In comparison, the untreated controls averaged 87,500 (Untreated Control 1) to 131,700 (Untreated Control 2) cells/mL (Fig. 3). Planktonic cyanobacteria were not detectable in negative controls (quartz sand and filtered site water). There were no significant differences in planktonic cell densities among the 3 granular algaecides at the highest concentration evaluated (P = 0.094). For GreenClean Liquid 5.0, average planktonic cell densities increased with increasing dose, but were not significantly different from each other (P = 0.81; Fig. 3). Although not statistically significant, average planktonic cell densities declined after treatments of all 3 granular formulations when applied at the maximum allowable treatment concentration, relative to the lower treatment concentrations and the untreated control. In addition, these effects were observed at 10 mg/L as H₂O₂ for the granular formulations, whereas the maximum concentration of the liquid formulation evaluated was 22 mg/L as H₂O₂. These lines of evidence supported our conclusion that the granular formulations were more effective than the liquid formulation in this experiment.

Figure 3. Average planktonic cyanobacteria cell densities (n = 3) following 18 d of incubation after laboratory algaecide experiments for single (2 graphs under A) and 2 hydrogen peroxide-based algaecide treatments (2 graphs under B). Granular sodium carbonate peroxyhydrate formulations are on left-hand side and liquid hydrogen peroxide formulation is on right-hand side. Error bars indicate ±1 SD.

Regarding treatment frequency, 2 of the granular algaecides yielded significant differences between single-dose and double-dose treatments (applied 48 h apart) when applied at 7 mg/L as H₂O₂. For example, when Phycomycin SCP was applied at 7 mg/L as H₂O₂, the average planktonic cell density was 60,000 cells/mL following one dose and <2500 cells/mL (detection limit) following 2 doses (P = 0.0083). For GreenClean Pro applied at 7 mg/L as H₂O₂, the average planktonic cell density was 199,200 cells/mL following one dose and 35,000 cells/mL following 2 doses (P = 0.026). There were no statistically significant differences between one dose and 2 doses for any of the granular algaecides when applied at 10 mg/L as H₂O₂ (GreenClean Pro: P = 0.7479; PAK 27: P = 0.41; Phycomycin SCP: P = 0.9971). Following treatments with Phycomycin SCP at 10 mg/L as H₂O₂, the average planktonic cell density was 8750 cells/mL following one dose and 2500 cells/mL following 2 doses. For GreenClean Pro, the average planktonic cell density was 47,500 cells/mL following one dose and nondetect (i.e., <2500 cells/mL) following 2 doses. There were no significant differences between one dose and 2 doses of PAK 27 at either concentration evaluated (7 mg/L as H₂O₂: P = 0.99; 10 mg/L as H₂O₂: P = 0.41). At the highest concentration of PAK 27 evaluated (10 mg/L as H₂O₂), average cell densities increased from 50,000 cells/mL following one dose to 340,000 cells/mL following 2 doses (Fig. 3).

Based on these results, we concluded that the granular SCP-based formulations were more effective than the liquid H₂O₂-based formulation but were all similar in efficacy when applied once at the maximum label concentration. We also concluded that 2 consecutive doses were more effective than one dose for 2 of the algaecides evaluated. GreenClean Pro was selected for the field application at a frequency of 2 treatments (48 h apart). The maximum allowable treatment concentration per this algaecide label is 10 mg/L as H₂O₂ and the algaecide is 27.6% peroxide by mass. The first treatment was applied at a concentration of 10 mg/L as H₂O₂ (36.8 kg algaecide/dam³, or 36.8 mg algaecide/L) and the second treatment was applied at a concentration of 7.8 mg/L as H₂O₂ (28.4 kg algaecide/dam³, or 28.4 mg algaecide/L). The follow-up treatment concentration was lowered slightly to decrease total costs, since laboratory data did not indicate significant differences between the 7 and 10 mg/L treatments when applied twice (P = 0.87).

Performance monitoring: immediate pre- and post-treatment sample analysis

In field samples collected 24 h before the first in-lake algaecide treatment in April, surface water samples contained relatively low cell densities of planktonic algae, with assemblages containing a mixture of diatoms, green algae, and cyanobacteria. Average total cell densities of planktonic cyanobacteria ranged from 800 cells/mL in the control zone to 1000 cells/mL in the treatment zone (Fig. 4C). Dominant cyanobacterial genera in assemblages included Aphanizomenon and Microcystis. Average resting cell densities in sediments were 14,700 and 182,400 cells/g in treatment and control zones, respectively (Fig. 4A). From these samples, average planktonic cell densities of 111,600 and 58,800 cells/mL, respectively, were measured following 14 d incubation studies in the laboratory (Fig. 4B). Thus, a relatively lower average cell density of resting cells in sediments in the treatment zone resulted in higher recruitment, whereas a higher average cell density in sediments from the control zone had lesser recruitment. This trend was also observed in the February and March 2022 sampling events prior to treatments (Fig. 2A). Water temperatures measured 1 d before treatment averaged 13.1 C and 12.8 C at the water surface and sediment–water interface, respectively, in the treatment zone (supplement). Similarly, water temperatures were 13.6 C and 13.5 C at the surface and sediment–water interface, respectively, in the control zone (supplement).

Figure 4. Immediate pre- and post-treatment evaluation of average resting cell densities (n = 3) in site-collected sediments (A), average planktonic cell densities (n = 3) in overlying water following laboratory incubation studies using site sediments (B), and average cell densities (n = 3) of planktonic cyanobacteria in situ at time of sampling 1 d before in-lake treatment (pretreatment) and 3 d after first in-lake treatment (post-treatment) (C) at Milford Gathering Pond. Error bars indicate ±1 SD. Samples were collected from sites W1, W2, and W3 for treatment zone and E1, E2, and E3 (shown in Fig. 1) for control zone.

On 22 April 2022 (3 d after first treatment, 1 d after second treatment), average resting cell densities in sediment samples increased slightly in the treatment zone to 41,800 cells/g and decreased slightly in the control zone to 160,800 cells/g (Fig. 4A), contrary to expectation. In 14 d incubation studies, resulting planktonic cell densities were 477,100 and 14,900 cells/mL from treatment and control zone samples, respectively (Fig. 4B). As with preceding samples, a relatively lower average cell density of resting cells in sediments within the treatment zone resulted in higher recruitment, and a higher average cell density in sediments within the control zone resulted in lesser recruitment to the water. Average planktonic cell densities of cyanobacteria in site-collected water samples were 710 cells/mL in the treatment zone and 1200 cells/mL in the control zone (n = 3; Fig. 4C).

Performance monitoring: May through October

In post-treatment samples collected from the surface water on 11 May 2022, average cell densities of planktonic cyanobacteria were relatively low, ranging from 2100 to 3900 cells/mL in the control and treatment zones, respectively (Fig. 5), indicating that planktonic growth was still relatively low throughout the lake. At the following sampling event (25 May 2022), average cell densities were 109,800 cells/mL in the treatment zone and 302,100 cells/mL in the control zone. Microcystis was the dominant genus in these first 2 sampling events. Samples were collected again 6 d later, following notification that the on-site fish hatchery would be discharging water with known HABs from various ponds into the treatment zone. The goal was to collect another set of samples before the treatment zone was confounded with these inflows. Samples were also collected from the ponds that were to be discharged to enumerate cyanobacteria. Prior to these hatchery discharges, average planktonic cell densities were 84,000 cells/mL in the treatment zone and 173,500 cells/mL in the control zone. Following the discharges, densities were 6,961,700 and 1,201,600 cells/mL in treatment and control zones, respectively (Fig. 5), and were entirely composed of Microcystis. In samples collected from 2 hatchery ponds prior to discharge, cyanobacterial cell densities ranged from 1000 to 191,200 cells/mL and contained Dolichospermum, Microcystis, and Raphidiopsis. Despite the treatment zone being confounded by hatchery discharges, average cell densities trended lower than in the control zone for all remaining sampling events (Fig. 5).

Figure 5. Average cell densities of planktonic cyanobacteria (n = 4) in treatment and control zones in the months following in-lake treatments at Milford Gathering Pond. Dashed line indicates date when fish hatchery released water containing HAB into treatment zone. Error bars indicate ±1 SD. Samples were collected from sites W4, W5, W6, and W7 for treatment zone and E4, E5, E6, and E7 for control zone (Fig. 1).

Discussion

Site characterization

The criteria for determining whether Milford Gathering Pond was an appropriate candidate site for this evaluation were detection of resting cells in sediments and measurable viability based on extent of transfer from sediments to water (i.e., recruitment) in laboratory incubation studies. These criteria can be used when evaluating whether detected resting cells could be a primary driver for HABs at a given site, since presence alone does not correlate with viability (Calomeni et al. 2022). At the study site, resting cells were detected in all samples collected at all sampling events before treatments, and evidence of viability was clear from the incubation studies performed in the laboratory. However, recruitment potential differed between treatment and control zones of the lake in samples collected in February and March 2022 (1–2 months before treatment), and in samples collected immediately prior to and after treatments. In each of these cases, planktonic cell densities measured after 14 d incubation were higher in the treatment zone samples than in control zone samples, despite having comparable densities of sediment-associated resting cells. These differences could be due to the number of viable resting cells from source sediments, sediment characteristics, and/or nutrient content variation between the zones (Karlsson-Elfgren et al. 2004, Faithfull and Burns 2006, Calomeni et al. 2022). In sediment samples collected in September 2021, total phosphorus concentrations were 1410 and 5960 mg/kg in control and treatment zones, respectively (supplement). Thus, higher phosphorus content in the treatment zone sediments could correlate with greater extent of recruitment and reproduction within that basin. Surface water concentrations of total phosphorus and orthophosphate also trended higher in the treatment zone than in the control zone for most of the sampling events (supplement).

Laboratory algaecide efficacy experiments

In the laboratory-scale algaecide efficacy experiments, we observed that concentrations of 10 mg/L as H₂O₂ (for granular SCP algaecides) were necessary to achieve measurable declines in recruited cell densities as compared to untreated controls. Calomeni et al. (2023b) also observed that exposures of 10 mg/L as H₂O₂ from a granular peroxide-based algaecide resulted in a 72% decline in recruited cell densities using sediments from Milford Gathering Pond relative to untreated controls for single-dose treatments. This experiment was conducted at a separate laboratory providing additional confidence in the accuracy of these data. Chen et al. (2016) found that concentrations of 5 and 20 mg/L of H₂O₂ elicited significant adverse effects to resting cells, based on pigment concentrations measured in “recovery experiments” following treatments (similar to the incubation studies utilized in this experiment). However, Chen et al. (2016) did not evaluate treatment concentrations between 5 and 20 mg/L, so data for direct comparison to the treatment concentrations used in this study are limited. Jia et al. (2014) observed that 10 mg/L as H₂O₂ in combination with 1 mg/L rice straw significantly decreased recruited cell densities of cyanobacteria in treated enclosures within Lake Taihu (China). However, H₂O₂ and rice straw were not evaluated separately in the pilot experiment (Jia et al. 2014).

The granular algaecide formulations showed greater effectiveness than the liquid formulation in this study. Calomeni et al. (2023b) also observed greater effectiveness from a granular peroxide-based algaecide for control of overwintering cyanobacteria as opposed to liquid-based formulations in a similar laboratory-scale experiment. Single-dose treatments of a granular peroxide formulation resulted in a 72% decline (relative to untreated controls) in average planktonic cell density, whereas 2-dose treatments (24 h apart) resulted in a 91% decline (Calomeni et al. 2023b) when resting cells in sediments were exposed. Comparatively, single-dose treatments of 2 different liquid peroxide algaecides resulted in greater average cell densities than the untreated control (Calomeni et al. 2023b). It is unclear whether these differences are correlated with the physical state of the algaecides, due to differences in formulations. The granular algaecides were all the same formulation with an active ingredient of SCP, which dissociates into H₂O₂, whereas the liquid formulation contained active ingredients of H₂O₂ and peroxyacetic acid. The SCP formulation allows for slower release of H₂O₂, especially at cooler water temperatures relevant to this treatment tactic. The H₂O₂ in the liquid algaecide can be immediately “spent”; thus, activity can be lost more quickly. Activity of algaecides and resulting bioavailability to target cyanobacteria can also be modified by various sediment characteristics. For example, since hydrogen peroxide is an oxidant, organic matter in sediments will also be oxidized when these algaecides are applied to sediments. Further, some activity may be lost as granules sink through the water column when applied. Higher concentrations will likely be necessary in preventative algaecide treatments (as compared to planktonic treatments) to overcome this activity loss and ensure appropriate exposure to target resting cells in sediments.

Two consecutive doses, 48 h apart, showed greater performance than a single dose in our laboratory experiments. Doubling frequency of treatments fundamentally increases the total exposure of algaecide, thus increasing the probability of eliciting effects to resting cells sufficient to manifest in measurable declines in recruitment. Based on our results, performing 2 preventative treatments can increase probability of success, and this frequency is allowable (minimum of 48 h apart) per the labels for the algaecides evaluated in this study.

Algaecide treatments

Overall, the algaecide applications were completed without any unforeseen issues. During the week of the treatments there were moderate to high winds (e.g., 24–40 km/h), which occasionally caused algaecide granules to be pushed sideways from the treatment transects during application. Granular spreaders are still likely to result in the most even broadcast, but some additional attachments to help direct granules closer to the water surface during periods of higher winds will be considered for future use. Use of weighted drop hoses that carry a slurry to the sediment surface are not recommended for preventative algaecide treatments targeted for resting cells in sediments, since this will create narrow lines of algaecide across sediments and leave more surface area untreated. Since peroxide-based algaecides are contact algaecides, target organisms must come in contact with the exposure for a treatment to be effective.

The measured differences in recruitment between treatment and control zones are useful in interpreting performance results. Despite the greater recruitment potential in the treatment zone and the hatchery discharges containing cyanobacteria to that area in late May 2022, monitoring data showed lower average cell densities of cyanobacteria in the surface water of the treatment zone than in the control zone for all the monitoring events following the hatchery discharges. Therefore, performance of the preventative treatments may have been greater than initially perceived based on monitoring data alone.

An additional challenge in interpreting performance data was the high variability within and among sample sites. Spatial heterogeneity of resting cells has been documented in prior studies (i.e., between shallow and deep areas; Cirés et al. 2013, Legrand et al. 2017). Therefore, future efforts will focus on improving sampling protocols to ensure sufficient statistical resolution to discern differences in treatment and control areas. This may include increasing the number of subsamples for each sample or increasing the total sample size to increase statistical power. Future studies will also aim for comparable sediment characteristics and resting cell viability in treatment and control zones to minimize confounding factors.

Preventative algaecide treatments were the only form of HAB management implemented at Milford Gathering Pond in 2022. To increase probability of achieving management goals, proactive algaecide treatments can also be included in management plans. This would include developing a strategic monitoring plan with a defined action threshold (e.g., cell density) to trigger algaecide treatments early in a growth cycle (Kinley-Baird et al. 2020). At this site, we do not anticipate that preventative treatments will eliminate the need for in-season management due to the densities of resting cells measured and history of severe HABs. However, preventative treatments may offset the growth potential of a HAB, thus requiring less frequent treatments or lower volumes of algaecide used overall. If both preventative and proactive treatments successfully decline densities of HABs and are implemented routinely, then over time, the supply of viable resting cells in sediments should also decrease and thereby decrease the potential for HAB formation. An additional approach that could be combined with preventative algaecide treatments would be in situ and watershed-scale nutrient management to minimize bioavailability of nitrogen and/or phosphorus to resting cells in sediments and in the water column.

Conclusions

This study evaluated the efficacy of preventative peroxide-based algaecide treatments targeted at cyanobacterial resting cells in sediments of a central US lake historically impacted by HABs. Our primary performance metric was defined as a decline in cell densities of HAB-forming genera in the water column throughout the growing season. For 9 of 11 sampling events from May to October, average cell densities trended lower in the treatment zone than in the control zone (ranging from 6 to 97% lower). Given that earlier sample analysis indicated higher recruitment potential of overwintering cells in treatment zone sediments, lower planktonic cell densities in the treatment zone (than in the control zone) indicate potentially greater performance than what could be perceived from monitoring data alone. However, the treatment zone was confounded by discharges from the fish hatchery as of early June, limiting our ability to correlate monitoring data with treatment performance. We did not detect measurable differences in resting cell densities in sediments immediately following treatments, nor were there relevant differences measured in recruitment potential following laboratory incubation studies using field-collected sediments containing resting cells. While the water column monitoring data provided a line of evidence to show that preventative algaecide treatments could be effective at decreasing HAB severity during warmer months, additional studies will be necessary to bolster the database supporting use of this management approach. If effective, preventative algaecide treatments could lessen HAB severity in valuable freshwater resources, which could alleviate the need for more aggressive forms of management during warmer seasons and allow lakes to remain safe and operational.

Acknowledgments

The authors thank Kansas Department of Health and Environment (KDHE) current employees Shawn Weber and Ben Growcock and former employees Josh Cullum, Patrick Olson, Adam Blackwood, and Tristan Runyan for field support, along with Leila Maurmann and other KDHE laboratory staff for analytical support. We also thank the team of aquatic applicators who performed the algaecide treatments: Adam Charlton, Mike Johnson, Mike Whitacre, Travis Pennington, and Dillon McConn of Aquatic Control, Inc. We are especially grateful to Daric Schneidewind (Kansas Department of Wildlife and Parks) and Ken Wenger (USACE- NWK District, Milford Reservoir Office) for coordination with curtain installation, algaecide offloading and storage, and assistance with field treatments and sampling.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Kansas State Water Plan Fund for HAB Mitigation and by the US Army Corps of Engineers (USACE) Aquatic Nuisance Species Research Program (ANSRP) [funding account code U4384107 and AMSCO code 008284].