Evaluation of a Peroxide-Based Algaecide for Cyanobacteria Control: A Mesocosm Trial in Lake Okeechobee, FL, USA

Abstract

A 72 h small-scale trial was conducted in enclosed mesocosms in the Lake Okeechobee waterway to evaluate the effectiveness of a USEPA-registered peroxide-based algaecide (formulated as sodium carbonate peroxyhydrate) for controlling a natural cyanobacteria population. Mesocosms were initially subjected to either no algaecide or the maximum label rate of 10 mg H₂O₂·L⁻¹. A subset of mesocosms were then subjected to a sequential application of 5 mg H₂O₂·L⁻¹ at 48 h after initial treatment. Following application, peroxide concentrations rapidly decreased and were undetectable by 48 h. At 24 h after treatment, significant decreases in all biomass indicators were observed (compared to untreated mesocosms), including extracted chlorophyll a, microscopic counts (total phytoplankton and total cyanobacteria), and cyanobacteria-specific 16S rRNA gene copies by over 71%. Although peroxide treatment reduced cyanobacteria biomass, there was no change in overall community structure and the remaining population was still dominated by cyanobacteria (>90%). After 48 h exposure, some biomass recovered in single application mesocosms resulting in only a 32–45% reduction in biomass. Repeated peroxide dosing resulted in the greatest efficacy, which had a sustained (60–91%) decrease in all biomass indicators for the entire study. While a single application of the peroxide was effective in the first 24 h, a sequential treatment is likely necessary to sustain efficacy when using this approach to manage cyanobacteria in the field. Results of this study support that this peroxide-based algaecide is a strong candidate to continue with scalable field trials to assess its potential future utility for operational management programs in the Lake Okeechobee waterway.

1. Introduction

Increased nutrient loading, particularly in freshwater systems, has been linked with the emergence of dense cyanobacteria harmful algal blooms (cyanoHABs) that are known to have severe negative impacts on ecosystem services [1]. Shallow subtropical lakes that lack stratification periods are especially vulnerable to excessive nutrient loads, where their phytoplankton communities are often dominated by cyanobacteria [2,3]. With changes in environmental conditions (e.g., temperature, CO₂, eutrophication, land usage, etc.) and water-use dynamics (e.g., dynamic rainfall, residence time, etc.), cyanoHAB intensity, duration, and associated toxicity are predicted to increase [4,5,6,7]. Both public awareness and rising human and ecological health concerns over cyanoHABs will likely prompt the necessity for large-scale mitigation measures [8,9,10]. However, small-scale studies are fundamental to the successful transition of field-ready technologies to operational use.

Short- and long-term cyanobacteria management initiatives should be considered in concert to provide a comprehensive integrated water management response plan in any freshwater system. While long-term management strategies typically focus on reducing nutrient loading and land-use patterns, short-term strategies, such as algaecide applications, can provide immediate relief from devastating bloom events [11,12,13,14,15]. Algaecides and other scalable intervention strategies have the potential to rapidly restore critical water uses and protect human health and lake ecology from the detrimental effects of cyanoHABs and associated toxins. Thus, the targeted use of US Environmental Protection Agency (EPA)-registered, peroxide-based algaecides have a high potential for immediate implementation.

Several USEPA-registered peroxide-based algaecides have proven effective in mitigating cyanoHABs and can be widely used in surface waters [11,14]. From this pool of algaecides, sodium carbonate peroxyhydrate was selected for evaluation. Peroxide-based algaecides are ideal for treatment since they have demonstrated negligible toxicity to non-target species [16], are shown to select for beneficial phytoplankton [17,18,19,20], have low exposure duration requirements for efficacy [14], dissipate rapidly with no residual environmental impacts [21], and have no water-use restrictions (e.g., drinking, swimming, fishing, irrigation, livestock watering) [22]. Peroxide-based algaecides are potentially operationally feasible, scalable, and can be readily implemented to rapidly elicit control of impacted sites to restore water uses. Peroxide-based algaecide treatments have also demonstrated a rapid decrease in cyanotoxins [11,14,17,23]; possibly from either direct oxidation, which is enhanced under ultraviolet light [24,25,26], or natural dilution/microbial degradation processes. Mesocosm studies provide an opportunity to evaluate peroxide exposures under representative conditions in the field with the increased predictive ability to translate results to larger field treatments. Field conditions such as light intensity and nutrients have been shown to be a critical aspect of peroxide effectiveness on cyanobacteria [27,28]. Targeting algaecide treatments in prioritized areas can also rapidly offset the risk to human and ecosystem health. Prioritized sites may include high-use areas such as beaches, marinas, campgrounds, parks, areas adjacent to spillway gates and locks, and residential canals or ecologically-sensitive sites where high-density cyanoHABs pose a clear risk to critical habitat and/or at-risk organisms.

The overall goal of this study was to evaluate a USEPA-registered peroxide-based algaecide in situ to mitigate naturally occurring cyanobacteria bloom in mesocosms in Lake Okeechobee. A future goal of this work is to increase the predictive ability of these types of products in operational field operations in shallow subtropical lakes. Specifically, the primary objectives included the following: (1) assess algaecide efficacy on cyanobacterial biomass through chlorophyll a and cell density, (2) determine the impact of the algaecide on cyanotoxin concentrations and associated genes, and (3) evaluate community-level changes following algaecide exposure.

2. Materials and Methods

2.1. Site Description and Study Design

The study was conducted at the City of Pahokee Marina (26°49′34.579″ N, 80°39′57.142″ W) on the southeastern shore of Lake Okeechobee located in Pahokee, Florida (USA). Water depth in the study area was approximately 2 m. This site was selected due to limited water exchange with the surrounding lake and the recent history of cyanoHABs (pers. obs.). Except for a small boat entrance, the marina is surrounded by a jetty that dampens wave action from the lake and reduces water exchange, making this an ideal demonstration site.

Twelve cylindrical limnocorrals were constructed from 6 mm-thick clear polyethylene tubing approximately 1 m wide and 1.7 m high (total volume ~1.34 m³) and sealed at the bottom to ensure the system was closed to lake water infiltration similar to Mitman and Tuccie [29] (Figure 1). Polyvinyl chloride (PVC) pipe (1.3 cm diameter) was used to build three support rings to aid in preventing limnocorral collapse. Limnocorrals were suspended using a foam-filled 10 cm corrugated flotation pipe (Figure 1) and were anchored to adjacent floating docks and to the marina bottom using 16-kg anchors. Limnocorrals were filled with the surrounding lake water that contained natural concentrations of cyanobacteria using a submersible pump. The tops of the filled limnocorrals, hereby termed mesocosms, were left open to allow for gas exchange and light penetration. Flotation pipes were high enough above the surface to prevent the introduction of surrounding water due to wave action. The study had a completely randomized design with four replications per treatment. Treatments consisted of untreated control, along with single (maximum label rate of 10 mg H₂O₂·L⁻¹) and sequential (maximum label rate of 10 mg H₂O₂·L⁻¹ followed by half maximum label rate of 5 mg H₂O₂·L⁻¹ 48 h after initial dose) applications of sodium carbonate peroxyhydrate (PAK^® 27; EPA Reg. No. 68660-9-67690, 2018; SePRO Corporation, Carmel, IN, USA). The amount of product applied was determined as follows: 1340 L (volume of mesocosms) × 10.2 mg H₂O₂·L⁻¹ (Max allowable rate per label) = 13,668 mg H₂O₂ needed/0.276 (product is 27.6% by mass H₂O₂) = 49.5 g product per mesocosm. Direct measurement of the corresponding amount of algaecide to achieve designated exposure concentrations was conducted using a calibrated balance (±0.001 g; Metler Toledo, Columbus, OH, USA). For treatment, algaecide was carefully hand spread over the entire mesocosm surface to ensure even distribution.

2.2. Sample Collection

Grab samples were collected using an integrated water sampler made from a 3.8 cm diameter PVC pipe with a rubber stopper to ensure a homogenized sample was collected from the surface to 1 m depth. The collected sample was used for all endpoints, with the exception of the in situ data sonde measurements. Collection times are noted in Section 2.3 for each endpoint.

2.3. In Situ Measurement Methods

In situ water quality parameters were assessed using a hand-held multi-parameter meter ProDSS sonde (YSI Inc., Yellow Springs, OH, USA) equipped with optical dissolved oxygen (DO), conductivity/temperature, and total algae sensors. Water quality data were collected from each mesocosm at 0, 6, 24, 48, and 72 h after initial treatment. Vertical profiles were collected in the center of each mesocosm over a 1 m depth profile with data collected in continuous mode and reported as an integrated median over the entire profile. The following parameters relevant to this study were collected: temperature (°C), specific conductance (SPC; mS·cm⁻¹), chlorophyll a (chl a; mg·L⁻¹), and DO (mg·L⁻¹). Data were retrieved using the software KorDSS v 1.6.6.0 (YSI Inc., Yellow Springs, OH, USA) and exported to Prism8 v 7.04 (GraphPad, San Diego, CA, USA) for analysis.

2.4. Peroxide Monitoring

Hydrogen peroxide (H₂O₂) concentrations were measured spectrophotometrically using the triiodide (I₃) method [30,31]. This method has been used to measure H₂O₂ in field treatments of sodium carbonate peroxyhydrate [12]. A standard curve was prepared using 30% H₂O₂ (Fisher Scientific, Waltham, MA, USA) (MDL = 0.3 mg H₂O₂·L⁻¹). Triiodide absorbance was measured using a Pasco PS-2600 spectrometer (Pasco Scientific, Roseville, CA, USA) following acidification reactions. Analyses were conducted on the homogenized integrated sample for all mesocosms at 0.17, 2, 4, 6, 24, and 48 h after initial algaecide application, along with 0.17, 4, and 24 h after the sequential algaecide application. Analyses were conducted on-site within 1 h of sample collection.

2.5. Extracted Chlorophyll

Samples were collected before treatment (0 h) and at 6, 24, 48, and 72 h post initial algaecide application. Grab samples were pre-processed on-site and frozen at −20 °C for analysis at the US Army Engineer Research and Development Center (ERDC) in Vicksburg, MS. Chlorophyll pigments were extracted according to the EPA method 446 in 90% acetone [32]. A modification included homogenization of filters by bead beating using DNase/RNase free 0.1 mm silica beads (lysing matrix B) at 4.5 m·s⁻¹ (~500 rpm) for 1 min using a Fast Prep 24 Homogenizer (MP Biomedicals LLC, Santa Ana, CA, USA). A total volume of 6 mL was used for pigment extractions. Absorbance (Abs) was measured on a UV-Vis 1800 spectrophotometer (Shimadzu, Kyoto, Japan) at 750 nm (turbidity) and 664 nm (chl a) for uncorrected chl a determination as reported in EPA method 446 [32].

2.6. Molecular Methods

Quantitative real-time polymerase chain reaction (PCR) was used to rapidly assess total cyanobacteria and their toxin-producing genes at 0, 24, 48, and 72 h post initial algaecide treatment using the CyanoDTec kit (Phytoxigene, Akron, OH, USA). Cyanotoxin gene presence/absence can be indicative of the potential for a bloom to produce toxins [33]. The CyanoDTec assay targets cyanobacteria specific 16S rRNA for total cyanobacteria and mcyE/ndaF for microcystins/nodularins, sxtA for saxitoxin, and cyrA for cylindrospermopsin. Briefly, water samples were filtered on-site onto 0.45 mm mixed cellulose ester (MCE) membranes and preserved at −80 °C (dry ice) until DNA extraction at the ERDC laboratory. Each filter was then homogenized using TRIzol Reagent (Fisher Scientific) with sequential freeze-thaw cycles in liquid nitrogen. Next, the homogenate was centrifuged and supernatant transferred to a fresh tube. To each homogenized field sample, 3 M sodium acetate was added at 10% the volume of the supernatant followed by ethanol precipitation. Precipitated DNA was washed twice with 70% ethanol and dissolved in Tris-EDTA (TE) buffer (Fisher Scientific). The genomic DNA concentration was then determined using a NanoDrop One microvolume UV-Vis spectrophotometer (Fisher Scientific). For short (days) and long-term (weeks) storage, extracted DNA was stored at 4 °C and −80 °C, respectively. Once extracted, all four replicates were concatenated at the same concentrations into 1 sample for qPCR screening via the CyanoDTec assay (Phytoxigene).

2.7. Microscopic Identification and Enumeration

Water samples were collected for qualitative and quantitative analyses of the phytoplankton community structure before treatment (0 h) and at 24 h and 72 h post initial algaecide application. Note that this analysis was not completed at 24 h for the sequential application mesocosms, as it was presumed to be similar to the single application mesocosms at this time, having both received the same single peroxide treatment. Each sample was preserved in the field with 4% formaldehyde (v/v), stored at 4 °C. Once in the lab (Ft. Lauderdale REC, University of Florida/IFAS), an acetic Lugol solution of 0.5% (v/v) was added to sediment cells. Sample bottles were mixed and a subsample was poured into Utermöhl chambers and sedimented overnight before counting [34]. Quantitative analyses were carried out using the Utermöhl method in 20 mL chambers [35], counting at least 20 random fields of view (using both vertical and longitudinal transects), with an inverted phase-contrast microscope (Olympus CKX41) [36]. Phytoplankton were identified using specialized literature [37,38,39,40,41,42] following higher classification in Guiry and Guiry [43]. Cell concentrations were calculated following Pappas and Stoermer [36] and APHA 10,200 F. [44].

2.8. Total Toxin

Total microcystin and saxitoxin were determined at 0, 24, 48, and 72 h post initial treatment using an ADDA-microcystin enzyme-linked immunosorbent assay (ELISA) and saxitoxins (PSP) ELISA, respectively (Eurofins Abraxis, Warminster, PA, USA). The bulk sample was frozen on-site at −20 °C for later analysis at the ERDC laboratory. The sample was homogenized via bead beating using 0.1 mm silica beads at 4.0 m·s⁻¹ for 1 min using a Fast Prep 24 Homogenizer (MP Biomedicals LLC). An aliquot of the sample homogenate was then used for the microcystins/nodularins or saxitoxin ELISAs (Eurofins Abraxis). Microcystin ELISAs were conducted according to EPA method 546 [45], with a limit of detection (LoD) of 0.10 µg·L⁻¹. Saxitoxin ELISAs were conducted according to the manufacturer’s instructions, with an LoD of 0.015 µg·L⁻¹.

2.9. Statistical Analyses

Where appropriate, statistical analyses were performed using the program Prism v.8.0 (GraphPad Software, San Diego, CA, USA). A two-way repeated-measures analysis of variance (ANOVA) was used to evaluate differences in DO, temperature, SPC, extracted chl a, and phytoplankton densities by treatment (untreated, single, and sequential applications) and time. A two-way mixed model ANOVA was used to evaluate differences in phytoplankton and cyanobacteria densities by treatment (untreated, single, and sequential applications) and time as there were missing samples given that microscopy was not undertaken on the 24 h sequential application mesocosms as it was presumed to be similar to the single application mesocosms at this time. For both ANOVAs, a Tukey’s Post-Hoc Test was used for multiple comparisons. An alpha of 0.05 was used to determine statistical significance. No statistical analyses were conducted on qPCR data owing to the analysis of concatenated triplicates into a single sample. The program PRIMER (Plymouth Routines in Multivariate Ecological Research, Albany, Auckland, New Zealand) version 7 (PRIMER-e) was used to assess relative changes in the total phytoplankton communities (microscopic data) by treatment (untreated, single, and sequential applications) and time (0, 24, and 72 h), data were analyzed via permutational ANOVA (PERMANOVA) using a similarity/resemblance matrix constructed from Bray Curtis dissimilarities (standard for ecological abundance data) following standardization (by total) and square root transformations. Comparisons were also made using a Shannon Diversity Index to account for both community richness (H’) and evenness (J’) in PRIMER where higher indices indicate more diverse (H’) and/or more evenly distributed (J’) communities. A two-way mixed-model ANOVA was then used to compare differences in H’ and J’ by treatment and time as it was conducted with the microscopy data. Statistical analyses were not conducted on toxin data as the majority of data was low or below the LoD.

3. Results

3.1. Peroxide Concentrations

Hydrogen peroxide (H₂O₂) was undetectable (<0.3 mg·L⁻¹) in all experimental mesocosms prior to treatment (Figure 2). For the first application (both single and sequential mesocosms), H₂O₂ concentration increased to an average of 10.28 ± 0.07 (standard error of the mean [SEM]) mg·L⁻¹ at 10 min after treatment and remained at approximately 10 mg H₂O₂·L⁻¹ for 4 h after treatment. After 4 h, H₂O₂ dissipated steadily according to a logarithmic decay and was non-detectable in all mesocosms by 48 h after initial treatment (measured prior to second application) (Figure 2). The half-life of the first treatment was 19 h. In the second treatment of 5 mg·L⁻¹, H₂O₂ concentration increased to 7.81 ± 0.51 (SEM) mg·L⁻¹ in sequential mesocosms 10 min after the application. However, H₂O₂ rapidly dissipated in these sequential application mesocosms to a non-detectable level by 24 h, with a half-life of <4 h (Figure 2).

3.2. Water Quality

The initial average DO concentration was 7.6 ± 0.04 (SEM) mg·L⁻¹ across all pretreatment mesocosms (n = 12) (Figure 3A). On average, the DO concentrations remained above 7.0 mg·L⁻¹ through the 72 h experiment for all test conditions. Untreated mesocosms had a significantly greater mean DO 6 h after the initial peroxide dosing (12.3 mg·L⁻¹, p < 0.05) when compared to all treated mesocosms (9.1 mg·L⁻¹). However, the mean DO in the untreated mesocosms declined to 7.7 mg·L⁻¹ by 24 h, similar to the single (7.9 mg·L⁻¹) and sequential (8.0 mg·L⁻¹) application mesocosms (p = 0.4519). Mesocosms that received a second application of algaecide at 48 h had a small but statistically significant (p < 0.05) elevation in their mean DO level at 72 h (8.1 mg·L⁻¹) when compared to the untreated mesocosms (7.1 mg·L⁻¹). There was no significant difference in DO concentrations between mesocosms treated with algaecide.

The capacity of water to hold oxygen is strongly associated with temperature [46]. The average initial temperature of all pretreatment mesocosms was 30.3 ± 0.08 (SEM) °C (n = 12) (Figure 3B). While no treatment effect was detected for temperature data (p = 0.1659), the mean temperature peaked at 33.7 °C 6 h after the initial peroxide dosing and ranged between 28.5 and 30.1 °C in all treatments for the remainder of the study (n = 4 per treatment).

Specific conductivity (SPC) provides an estimate of the amount of dissolved ionic material in the water, where higher SPC values indicate more dissolved free ions. This is indicative of leaky or lysed cells [47] and decaying organic matter. Mean SPC was 431.5 µS·cm⁻¹ across all pretreatment mesocosms at 0 h (n = 12) (Figure 3C) and was consistent within the untreated mesocosms throughout the study ranging from 422.0 to 424.0 µS·cm⁻¹ between 24 and 72 h (n = 4 per condition). At 6 and 24 h only, the mean SPC of the single application mesocosms (491.0 and 460.3 µS·cm⁻¹, respectively) was significantly greater (p < 0.05) than that of the untreated control mesocosms (424.0 and 423.6 µS·cm⁻¹, respectively). Mean SPC of the sequential application mesocosms was significantly greater than untreated controls across the study, ranging from 442.6 to 493.7 µS·cm⁻¹ (p < 0.05). Additionally, the averages of the single and sequential application mesocosms did not significantly differ from one another at any time point.

3.3. Biomass Assessment

3.3.1. Phytoplankton

Major visual differences in algal presence between treated and untreated mesocosms were evident within minutes of initial algaecide application (Figure 4). Observations indicated that mesocosms dosed with the algaecide rapidly lost pigmentation within 10 min and did not recover by 24 h (Figure 4B,D,F). Alternatively, untreated mesocosms, retained growing algal populations (Figure 4A,C,E).

Chlorophyll a is a common indicator of algal biomass. The average extracted chl a concentration before algaecide application was 49.3 ± 3.3 (SEM) µg·L⁻¹ (n = 12) (Figure 5A). In the untreated control mesocosms, the mean extracted chl a concentration declined from 30.2 µg·L⁻¹ at 6 h to 19.7 µg·L⁻¹ at 72 h. Extracted chl a was significantly (p < 0.05) higher in the untreated control mesocosms when compared to the single and sequential application mesocosms at all time points except for 72 h, where the untreated and single application treatment were not significantly different from each other (p = 0.6024). Both the single and sequential application treatments saw an initial rapid decrease in extracted chl a at 6 h (9.2 and 9.8 µg·L⁻¹, respectively); a 68–70% decrease in biomass compared to untreated controls at the same time. At 24 h chl a was at its lowest in treated mesocosms at 6.3 µg·L⁻¹ and 8.0 µg·L⁻¹ in single and sequential mesocosms, respectively, representing an 82% and 77% average decrease in biomass. At later timepoints (48 h and 72 h) there was a gradual increase in the average chl a concentration of the single application treatment to 10.2 µg·L⁻¹ (48 h) and 13.4 µg·L⁻¹ (72 h), still representing a 58% and 32% reduction in chl a compared to respective untreated controls. Alternatively, at later timepoints in sequential treatment mesocosms chl a concentrations remained low with an increase to 8.4 µg·L⁻¹ at 48 h and 8.0 µg·L⁻¹ at 72 h, representing a 65% and 60% reduction in extracted chl a content that was sustained throughout the study. At no time did the two algaecide treatment mesocosm types significantly differ from one another.

Prior to peroxide application (0 h), mean phytoplankton cell density across all mesocosms was 1.32 × 10⁵ cells·mL⁻¹ (n = 12), indicative of generally accepted bloom concentrations (2.0 × 10⁴ to 1.0 × 10⁵ cells·mL⁻¹) [48]. At 24 h, mean phytoplankton cell density in the untreated mesocosms at 2.57 × 10⁵ cells·mL⁻¹ was significantly greater (p < 0.05) than that of the single application mesocosms at 7.11 × 10⁴ cells·mL⁻¹ (Figure 5B), where the single application mesocosms had a 72% reduction in cell density. At 72 h, the mean cell density of the untreated mesocosms at 2.46 × 10⁵ cells·mL⁻¹ was not significantly different than the single application mesocosm at 1.37 × 10⁵ cells·mL⁻¹ (p = 0.0506) but was significantly different in sequential application mesocosms at 6.56 × 10⁴ cells·mL⁻¹ (p < 0.05), where there was a 44% and 73% reduction in cell density compared to untreated controls, respectively.

3.3.2. Cyanobacteria

In pretreatment mesocosms (0 h), average cyanobacterial 16S rRNA copy number ranged from 2500 to 4213 copies·mL⁻¹ (n = 3 [4 concatenated replicates for each]) (Figure 5C). In untreated controls (n = 1 [4 concatenated replicates for each]), gene copy number after 24, 48, and 72 h was 9,037, 17,316, and 21,644 copies·mL⁻¹, respectively. In single application mesocosms, 16S rRNA copy number decreased to 282 copies·mL⁻¹ at 24 h, representing a 97% decrease in copy number compared to untreated controls. By 48 h the single application mesocosm started to recover with 4,507 copies·mL⁻¹ but still having 74% fewer gene copies than untreated controls. By 72 h single application mesocosms had recovered even further to 12,898 copies·mL⁻¹ with an overall 40% reduction in gene copies when compared to untreated controls at the same time. Lastly, in sequential application mesocosms 16S rRNA copy number decreased to 1,422 copies·mL⁻¹ at 24 h and 3,032 copies·mL⁻¹ at 48 h, having 84% and 82% decreases in copy numbers compared to untreated controls. At 72 h, 16S rRNA copy number continued to decrease in sequential mesocosms to 1920 copies·mL⁻¹ resulting in a 91% decrease in gene copy number.

Mean cyanobacteria cell density confirmed visual observations and qPCR results for cyanobacteria biomass. Prior to peroxide application, mean cyanobacteria cell density across all mesocosms was 1.19 × 10⁵ cells·mL⁻¹ (n = 12) (Figure 5D), accounting for 87% to 92% of the total phytoplankton population. In untreated mesocosms, mean cyanobacteria density increased to 2.46 × 10⁵ cells·mL⁻¹ at 24 h and remained relatively constant at 2.39 × 10⁵ cells·mL⁻¹ at 72 h. In single application mesocosms, there was a significant decrease in cyanobacteria concentrations to 7.18 × 10⁴ cells·mL⁻¹ at 24 h (p < 0.05) but some recovery was observed at 72 h with 1.32 × 10⁵ cells·mL⁻¹ (p = 0.0526), representing an initial 71% (24 h) and final 45% (72 h) reduction in cyanobacteria concentrations (Figure 5D). Additionally, the mean cyanobacteria density of the untreated mesocosms was significantly greater (p < 0.05) than that of the sequential application mesocosms at 6.03 × 10⁴ cells·mL⁻¹ at 72 h (n = 4). Although there were substantial decreases in cyanobacteria abundance at 72 h post-treatment, cyanobacteria still dominated the phytoplankton community at 91% to 97% across all conditions.

3.4. Community Level Changes

For the Shannon Diversity Index, genus-level mean community diversity (H’) ranged from 2.270 to 2.528 across all mesocosms at each timepoint (

Table 1. Shannon Diversity Index results highlight genus-level community diversity and evenness before and after algaecide exposure.

	Diversity (H′)			Evenness (J′)
Time	Untreated	Single	Sequential	Untreated	Single	Sequential
Pretreatment	2.270	2.386	2.415	0.8150	0.8510	0.8425
24 h	2.528	2.464	NA	0.8311	0.8473	NA
72 h	2.401	2.406	2.457	0.8088	0.8179	0.8446
NA indicates no analysis was conducted

). Additionally, mean community evenness (J’) ranged from 0.8088 to 0.8510 across all mesocosms and time points (Table 1). There was no statistically significant difference between any treatment at any time point for community diversity (0.3619 ≤ p ≤ 0.9999) or evenness (0.1596 ≤ p ≤ 0.9999) suggesting there was no shift in overall community structure. Results of the

PERMANOVA also supported that there was no change in community structure with or without the algaecide on standardized cell concentrations (0.1029 ≤ p ≤ 0.5742). Collectively, these data suggest that community composition was not impacted by algaecide treatment as no major changes in community structure could be directly linked to treatment condition, even though peroxide treatment resulted in an overall significant decrease in total phytoplankton/cyanobacteria.

Although there was no difference in community composition for untreated and treated mesocosms (single and sequential) there was a small change in the relative proportion of major cyanobacteria genera in the population over time. Pretreatment mesocosms (average) were dominated by members of Pseudanabaena (32%), Planktolyngbya (28%), Merismopedia (15%), Aphanocapsa (12%), and Microcystis (11%) (Figure 6). At 24 h, mesocosms (average, independent of treatment) were dominated by Pseudanabaena (31%), Aphanocapsa (29%), Planktolyngbya (13%), and Merismopedia (10%). At 72 h, mesocosms (average, independent of treatment) were also dominated by Pseudanabaena (35%) followed by Aphanocapsa (23%), Planktolyngbya (19%), and Merismopedia (16%). Major changes included an increase in Aphanocapsa from 12% to 23% at 72 h post-treatment and a decrease in Microcystis from 11% to <1% at the end of the 72 h study. Collectively, this suggests that more sensitive cyanobacteria could have been impacted by mesocosm artifacts but that these impacts did not disproportionately impact the algaecide treatments.

Although the phytoplankton community was dominated (>91%) by cyanobacteria before and after the algaecide treatment, the rest of the phytoplankton community was highly diverse. Initially, the average non-cyanobacteria community (pretreatment mesocosms) was dominated by Chlorodendrophyceae (50%), Dinophyceae (20%), and Chlorophyceae (11%), and the remaining phytoplankton groups comprised less than 10% of the non-cyanobacteria community (Figure 7). At 24 h, mesocosms (average independent of treatment) were dominated by Chlorophyceae (29%), Trebouxiophyceae (12%), and Coscinodiscophyceae (10%). Additionally, in the control mesocosm at 24 h Euglenophyceae as represented a substantial part of the community (34%). At 72 h mesocosms (average, independent of treatment) were dominated by Chlorophyceae (48%) followed by Trebouxiophyceae (8%) and Coscinodiscophyceae (8%) to a lesser extent. At 72 h the sequential mesocosms had a high proportion of Bacillariophyceae at 40% of the community. In general, there was no clear impact of the algaecide that could be separated from any potential mesocosm artifacts.

3.5. Cyanotoxins

For the microcystin gene (mcyE/ndaF), average composite copy number in pretreatment mesocosms was 235 copies·mL⁻¹ (n = 3 [4 concatenated replicates for each]). In untreated controls, mcyE/ndaF copy number increased from 3512 copies·mL⁻¹ at 24 h to 7935 and 14,418 copies·mL⁻¹ at 48 h and 72 h, respectively. In single and sequential mesocosms the mcyE/ndaF gene concentration was undetectable as the Ct values were more than four Ct values above the lowest standard (recommended by the manufacturer) at all time points evaluated. Additionally, microcystins/nodularins ELISA revealed concentrations below 1.07 ± 0.26 (SEM) µg·L⁻¹ for all mesocosms independent of treatment or time.

For the saxitoxin gene (sxtA), average composite copy number was below the level of detection in pretreatment mesocosms. In untreated control samples at 24 h there were 153 copies·mL⁻¹ that increased to 2755 copies·mL⁻¹ and 3037 copies·mL⁻¹ at 48 h and 72 h, respectively. Furthermore, in the single application mesocosms sxtA gene concentration increased from undetectable to 1068 copies·mL⁻¹ and 3333 copies·mL⁻¹ at 48 h and 72 h, respectively. Lastly, in the sequential mesocosm, only the 72 h timepoint was evaluated as the 24 h and 48 h results were presumed to be similar to the single application, at 72 h after the initial application there were 158 copies·mL⁻¹ of sxtA. Furthermore, saxitoxin ELISAs showed concentrations below 0.044 ± 0.006 (SEM) µg·L⁻¹ across all mesocosms and timepoints.

For the cylindrospermopsin gene (cyrA), no gene copies were detected at any timepoint for any treatment condition. Furthermore, no ELISAs were conducted as a major gene in the gene cluster responsible for producing cylindrospermopsin was lacking.

Given the low gene copy numbers and measured cyanotoxin concentrations that were observed, no link could be made with exposure to the algaecide for microcystins/nodularins, saxitoxins, or cylindrospermopsin for this bloom event.

4. Discussion

Peroxide-based algaecides (registered by the USEPA) have the potential to provide short-term rapid cyanobacteria control strategies in freshwater systems. However, mesocosm investigation field trials, as those in the current study, are necessary to assess product efficacy prior to the transition of scalable technologies to management programs for routine operational use; in part, due to differential cyanobacteria responses and environmental conditions (e.g., hardness, turbidity, light, etc.) across lakes, rivers, and reservoirs. In this study, a USEPA-registered peroxide-based algaecide was used to control a natural cyanobacteria bloom in enclosed mesocosms in the Lake Okeechobee waterway to evaluate its potential future use in this system. Overall, this product caused a significant decline in cyanobacteria biomass as determined by chlorophyll a, cell counts, and 16S rRNA cyanobacteria gene copies. The highest algaecide efficacy was observed under two sequential peroxide doses showing lasting efficacy for at least 72 h (the duration of this study). Furthermore, there were no major shifts in the phytoplankton community and only minor shifts in the cyanobacteria community. As such, this study supports the use of a USEPA-registered peroxide-based algaecide for the short-term control of natural cyanoHABs within Lake Okeechobee. This product could be of particular use for at-risk areas including high-use areas like beaches, marinas and parks, locks, and ecologically-sensitive sites in the Lake Okeechobee waterway.

Concerns exist surrounding the use of chemical management of nuisance species due to the putative risk of harming non-target organisms including algae, macroinvertebrates, and fish [49,50]. Many of these concerns have arisen due to the use of copper-based algaecidal products [49,50]. For example, copper sulfate is effective in managing cyanoHABs but can be toxic to some fish and water flea species under certain environmental conditions related to aqueous pH [49,50]. However, peroxide-based algaecides likely have a higher margin of safety than other chemical algaecides for most non-target aquatic organisms [16,51,52] as they are short-lived (minutes to hours), and have benign breakdown products of water and oxygen [21]. For example, Geer et al., [52] investigated the non-target toxicity of sodium carbonate peroxyhydrate (Phycomycin^®) to several freshwater organisms including a green alga (Pseudokirchneriella subcapitata), zooplankton (Ceriodaphnia dubia), amphipod (Hyalella azteca), and fathead minnow (Pimephales promelas). Geer et al., [52] showed 96 h LC50 concentrations ranging from 1.0–19.7 mg H₂O₂·L⁻¹ where the zooplankton was the most sensitive and fathead minnow the least sensitive. Based on current registered algicide application rates of SCP formulations (max ~10 mg H₂O₂·L⁻¹) the fathead minnow is unlikely to be adversely affected. While there is little to no margin of safety for peroxide-based algicides for the other organisms tested, there is evidence that peroxide-based algicides may have an enhanced margin of safety when compared to copper-based algicides, endothall, and diquat dibromide [52]. Furthermore, while the maximum application rate was used here, cyanobacteria control can likely be achieved at lower application rates, for example, Geer et al., [52] reported a 7-day EC50 for M. areuginosa of 0.9–1.0 mg H₂O₂·L⁻¹. In this study, a rapid decline in peroxide concentration was observed within the first 48 h, having a half-life of 19 h (Figure 2). As the mesocosms were sealed from water exchange within the lake, the observed decrease in peroxide concentration within the first 48 h was most likely from the uptake of peroxide by cyanobacteria and natural degradation rather than dilution. Matthijs et al., [11] similarly found a rapid decline in peroxide concentration to undetectable levels by 48 h after treatment. While the effects of the algaecide on non-target species (from phytoplankton to fish) were not directly assessed in the current study, the observed rapid decline in peroxide concentration suggests that non-target organisms are at a relatively low risk of exposure to concentrations that may have adverse effects. Furthermore, the non-target impacts of taking no action against a cyanobacteria bloom likely outweigh any non-target impacts of peroxide-based algaecides.

In water, peroxides form reactive oxygen species that damage DNA and membranes (e.g., cell wall, thylakoid), as well as oxidize proteins and lipids. In general, cyanobacteria are more sensitive to peroxide exposures than eukaryotic microalgae [53,54,55]. This could be due to more exposed phycobilisomes (not membrane-bound) or lack of enzymes that could ameliorate toxic effects [21,56]. The effectiveness of the peroxide-based algaecide against cyanobacteria was visually evident within 10 min, as seen in Figure 4, and mesocosms dosed with peroxide had still not visually recovered by 24 h post-treatment when compared to the untreated controls. Additionally, slight differences in DO and SPC between untreated and treated mesocosms within the first 24 h (Figure 3) also confirmed the effectiveness of the peroxide-based algaecide against this population of cyanobacteria, as these changes demonstrate a decrease in primary productivity and an increase in cell lysis. Additionally, observed DO levels were sufficient to maintain aquatic organisms important to preserving healthy aquatic ecosystems. Furthermore, biomass indicators including chlorophyll a, microscopic counts (total phytoplankton and total cyanobacteria), and 16S rRNA gene copies specific to cyanobacteria all showed a significant decline in concentrations after 24 h in all treated mesocosms (Figure 5). After 24 h some recovery was observed in mesocosms that only received one peroxide dose and by 72 h there was no difference between the untreated and single-dose mesocosms. This recovery however was not observed in mesocosms that received a second dose of the algaecide 48 h apart. While a single maximum application of the peroxide-based algaecide was effective initially, a sequential treatment may be required to sustain efficacy following an initial application when using this approach to manage cyanoHABs in the field.

The selective control of cyanobacteria species and community shifts away from cyanobacteria have been observed in other works [17,18,23]. In this study, while the overall density of the phytoplankton community was significantly reduced with the algaecide treatment, the remaining phytoplankton and cyanobacteria community structures were largely unchanged (Table 1) and the communities were still dominated by cyanobacteria (Figure 5B,D). As such, the peroxide application likely had a minimal impact on non-target phytoplankton since they represented a relatively small fraction of the total phytoplankton community. However, the cyanobacteria remaining after initial and sequential peroxide applications could serve as a source for future blooms, so monitoring of the population would still be needed, particularly if concerned about a succession event. Routine applications are often needed due to re-growth or if substantial biomass is initially present [11,57]. Though shifts in community structure were not observed in this short study, a rapid decline in peroxide concentration, especially in the presence of organic matter (i.e., algal biomass), could allow cyanobacterial genera that were initially present at low levels more opportunity to rapidly grow. This could happen through the lysis of a subset of the cyanobacteria population and subsequent conversion of organic matter (by heterotrophic bacteria) to dissolved organic matter for re-uptake by unaffected/less-sensitive cyanobacteria [58,59]. Additional longer-term studies are needed to assess overall phytoplankton community impacts from prolonged and repeated exposure to peroxide-based algaecides. Ideally, the efficacy of any algaecide should be tested against specific target species prior to community-level testing. Algaecide screening should be conducted in confined field trials prior to large-scale field applications to ensure biomass reductions and avoid the potential for a more toxic succession event, owing to presumed differential sensitivities among species that have previously been reported [17,18,23]. In situ mesocosm screening on natural populations as conducted here offers the potential for more realistic efficacy evaluations. It also provides information on community-level effects that would be difficult to achieve in laboratory evaluations for peroxide-based algaecides, particularly given the documented positive interaction between natural sunlight and the mode of action of this class of algaecides [60].

Overall, the peroxide-based algaecide tested was effective in reducing cyanobacteria density within 24 h of initial application. Sequential applications of peroxide will likely be needed to keep the cyanobacteria community at low densities, but likely can be applied at lower rates. Given how quickly peroxide degraded in the water column, even repeated applications of peroxide-based algaecides should have limited effects on water quality and non-target species. The observed effectiveness of the peroxide-based algaecide PAK^® 27 on reducing cyanobacteria density merits larger-scale management trials to standardize required dosage and application protocols under various environmental conditions. Furthermore, it also warrants additional small-scale screening studies on various genera of cyanobacteria to avoid differential control of mixed cyanobacteria assemblages that could lead to potentially more toxic, successive bloom events. Additionally, the potential to use peroxide-based algaecides in conjunction with other algaecides should be explored if they alone cannot effectively manage all target species of cyanobacteria. The results of this study suggest that this peroxide-based algaecide is a strong candidate to continue with larger-scale field trials to assess its use in routine management programs in the Lake Okeechobee waterway. For example, studies are needed on community recovery and resistance in natural cyanobacteria populations in addition to optimization studies on algaecide dosage and timing for prolonged efficacy to ensure that the implementation of this control strategy is successful in future management programs.