ASABE Logo

Article Request Page ASABE Journal Article

Microcystin Shows Thresholds and Hierarchical Structure With Physicochemical Properties at Lake Fayetteville, Arkansas, May Through September 2020

Brian E. Haggard1,*, Erin Grantz1, Brad J. Austin1, Abbie L. Lasater2, Lillie Haddock2, Alyssa Ferri3, Nicole Wagner4, J. Thad Scott5


Published in Journal of the ASABE 66(2): 307-317 (doi: 10.13031/ja.15273). Copyright 2023 American Society of Agricultural and Biological Engineers.


1Arkansas Water Resources Center, University of Arkansas, Fayetteville, Arkansas, USA.

2Biological and Agricultural Engineering, University of Arkansas, Fayetteville, Arkansas, USA.

3Crop, Soil and Environmental Sciences, University of Arkansas, Fayetteville, Arkansas, USA.

4Center for Reservoir and Aquatic System Research, Baylor University, Waco, Texas, USA.

5Biology, Baylor University, Waco, Texas, USA.

6Biological Science, Oakland University, Rochester, Michigan, USA.

*Correspondence: haggard@uark.edu

The authors have paid for open access for this article. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License https://creative commons.org/licenses/by-nc-nd/4.0/

Submitted for review on 18 July 2022 as manuscript number NRES 15273; approved for publication as a Research Article by Associate Editor Dr. Durelle Scott and Community Editor Dr. Kati Migliaccio of the Natural Resources & Environmental Systems Community of ASABE on 21 December 2022.

Highlights

Abstract. Harmful algal blooms (HABs) in freshwaters are a global concern, and research has focused on the nutrient drivers of cyanobacterial growth and toxin production. We explored the importance of nutrients on sustained cyanobacterial HABs producing measurable microcystin at Lake Fayetteville, Arkansas, USA. The specific objectives were to (1) quantify sediment phosphorus (P) flux and estimate potential equilibrium P concentrations (EPC0) in July 2020, (2) assess water quality conditions in the lake from March through September 2020, and (3) evaluate physicochemical thresholds (or change points, CPs) and hierarchical structure with total microcystin concentrations. The sediments were a potential P source under both oxic and anoxic conditions, and the SRP concentrations in the lake water were continuously less than the EPC0 estimated for bottom sediment (~0.03 mg L-1); sediments are likely a potential P source for cyanobacterial HABs at Lake Fayetteville. The physicochemical changes at Lake Fayetteville over the 2020 growing season were typical of small, hypereutrophic reservoirs, with low biomass in winter when nutrient supply was greatest and the greatest cyanobacterial growth and microcystin toxin as nutrient supply diminished into the growing season. Microcystin concentrations were elevated above 1 µg L-1 from mid-June through mid-August 2020, and most physicochemical parameters in this study showed thresholds or change points with microcystin. Hierarchical structure existed with total microcystin concentrations, showing the potential importance of cyanobacterial biomass, N supply, and total P on elevated microcystin. Nutrients and algal pigment raw fluorescence explained 83% of the variation in total microcystin concentrations at Lake Fayetteville during the 2020 growing season. Nutrients (both N and P) from external and internal sources are likely important drivers of these blooms and toxicity at Lake Fayetteville.

Keywords. Harmful Algal Blooms, Nutrient Drivers, Sediment Phosphorus Release, Water Quality.

Harmful algal blooms (HABs) in inland freshwaters and estuaries are an international issue and a human health threat (e.g., see Brooks et al., 2016), and if we look for algal toxins, particularly microcystin, then these toxins are often present in nutrient-rich freshwaters. For example, every state within the USA has had measurable concentrations of microcystin reported in multiple lakes over the last decade (EWG, 2020). Cyanobacteria toxins and HABs are widespread, and the mechanisms that support excessive algal growth and trigger toxin production have been and continue to be an important research focus across the USA and worldwide.

These HABs and nuisance algal blooms are associated with nutrient enrichment within the waterbody and its contributing watershed (Anderson et al., 2002; Paerl et al., 2016a; Wurtsbaugh et al., 2019). Often the nutrient sources resulting in HABs are external, such as agricultural runoff (Kane et al., 2014), septic systems (Lapointe et al., 2017), and effluent inputs (LaPoint et al., 2015). Once nutrients are in aquatic ecosystems, they cycle tightly between the water column and sediments, and internal nutrient loading can fuel HABs.

Phosphorus (P) from sediments can fuel algal growth and HABs in inland waters, especially when nitrogen (N) is not limiting (Orihel et al., 2015; Austin et al., 2020). Algal P demand can drive sediment P release in eutrophic waters (McCarty, 2019) when dissolved P in the water is less than concentrations in equilibrium with bottom sediments. This internal P source may result in seasonal N limitation of HABs (Ding et al., 2018), particularly when episodic N inputs are reduced during prolonged droughts. Nitrogen availability to HABs can either limit or trigger the production of toxins, including microcystin (Wagner et al., 2021). However, dual N and P management of both internal and external sources will likely be needed to control HABs in lakes and reservoirs and protect downstream waterbodies (Jankowiak et al., 2019; Paerl et al., 2016b).

If internal P sources are a potential threat to water quality and a HABs driver, then the first step is quantifying release rates from bottom sediments under oxic and anoxic conditions. In the summer of 2020, we estimated sediment P release rates in intact sediment-water cores from a hypereutrophic, recreational impoundment, Lake Fayetteville, in northwest Arkansas. The specific objectives were to assess HABs and water quality conditions during the growing season (i.e., March through September) at Lake Fayetteville and evaluate the physicochemical thresholds and hierarchical structure with total microcystin concentrations during an extensive cyanobacterial HAB. We hypothesized that hierarchical structure would account for more variation in microcystin and that physicochemical properties would exhibit thresholds with total microcystin concentrations. The goal was to understand the importance of physicochemical properties in the context of HAB management at a recreational lake. This analysis should be beneficial to the municipal authorities responsible for this recreational lake, providing the needed analysis to implement public health advisories when the potential for HABs and elevated toxins exist in the surface water.

Materials and Methods

Study Site Description

Lake Fayetteville (36.137092, -94.139794) is a surface impoundment in Washington County, northwest Arkansas, constructed in 1949, where cyanobacterial HABs have occurred, resulting in public advisories being issued. The lake has a maximum depth of up to 15 m and a surface area of ~0.6 km2, with a watershed of ~24 km2. The watershed land use represents primarily pasture-based agriculture (50%) and urban development (38%), based on NCLD 2019 using Model My Watershed. The contributing area has a long history of manure applications, poultry production, and land use change to reach the current proportions. The reservoir stores up to 90% of P input, while N retention is roughly 50% of input due to high denitrification rates (Grantz et al., 2014). In 2019, microcystin concentrations were measured up to 15 µg L-1 at the lake (Wagner et al., 2021), which was above the recommended guidelines for recreational contact waters (8 µg L-1, EPA, 2019). Since the Arkansas Water Resources Center (AWRC) began routine monitoring at this lake in December 2018, microcystin has been observed in measurable concentrations (i.e., greater than the reporting limit of 0.3 µg L-1) throughout the year (Wagner et al., 2021). Lake Fayetteville is an important recreational resource for the City of Fayetteville, including almost 80 ha of water surface and 185 ha of land around the lake (City of Fayetteville, 2021). No swimming is allowed at Lake Fayetteville; only secondary contact recreation is allowed, like boating, fishing, and kayaking.

Lake Sampling, Lab Methods, and Data Analysis

The AWRC collected water samples from just below the surface at three sites across the north shore of Lake Fayetteville on an approximately weekly basis (fig. 1). These sites were chosen because they represent public access points in the main park area. Water samples have been collected almost weekly since 2019, and data from March 2020 through September 2020 were presented and evaluated in this study, representing what we define as the 2020 growing season. The water samples were collected manually from boat docks or the shoreline, where public and pet interaction would possibly occur. We took care not to sample from water disturbed by the waves or wading into the waters. The sampling dates represented the range of hydrologic conditions across the study period, although we did not sample if conditions were adverse.

Water samples were analyzed for soluble reactive P (SRP), total P (TP), nitrate-N plus nitrite-N (hereafter, NN), soluble ammonia (NH4-N), total N (TN), and chlorophyll-a (Chl-a) at the certified AWRC water quality lab using standard methods (see: https://awrc.uark.edu/water-quality-lab/). Water samples were analyzed for total microcystin using an enzyme-linked immunosorbent assay (EPA Method 546; Abraxis, Warminster, PA) following three freeze-thaw cycles; the AWRC water quality lab participates in the proficiency testing program for recreational waters. Water samples were also analyzed for raw fluorescence units (RFUs) of chlorophyll (CHL) and phycocyanin (PC) using a CyanFlour (Turner Designs, San Jose, CA); these data suggest that cyanobacteria dominate the phytoplankton community when ratios approach and exceed one (Thomson-Laing et al., 2020).

We used these water quality data to evaluate physicochemical thresholds and hierarchical structure with microcystin concentrations (dependent variable) from March to September 2020; the data included all three sites (n=93). Nonparametric change point analysis (nCPA) was used to evaluate individual parameter thresholds, providing an estimate of the threshold and the 95% confidence interval about that estimate (Qian et al., 2003; King and Richardson, 2003). A categorical and regression tree (CART) (De’Ath and Fabricius, 2000) was used to determine thresholds and structure (i.e., additional splits or thresholds) in physicochemical data with total microcystin concentrations. We defined CART R2s as the deviance explained by each split relative to the total deviance (R. King, personal communication, 5 April 2022); this is used as R2 to show the percent of deviance explained. The nCPA and CART analyses were performed in R 4.1.2 using the rpart package for CART analysis (Therneau and Atkinson, 2019), which required at least five observations per terminal node and that each split in the model explained at least 5% of the dataset variance (i.e., complexity parameter = 0.05).

Figure 1. Sampling sites for routine lake monitoring at Lake Fayetteville, Arkansas, summer 2020, sediment cores were collected in deepest part of lake directly south of routine sites, and images show marina that was sampled (photos credit: B.J. Austin and B.E. Haggard).

Sediment Cores and Phosphorus Flux

In July 2020, a total of 21 sediment cores were collected at Lake Fayetteville from the main channel near the dam, mid-lake, and inlet using a UWITEC corer (fig. 1). Plexiglas tubes (0.6 m in length) were inserted approximately 0.3 m into the sediments, and then the cores and overlying water were capped on the bottom and stoppered on top. A properly collected sediment-water core had relatively undisturbed surface sediments and relatively clear overlying water. 

Upon return to the AWRC water quality lab, the depth of the water and sediments (if needed) were adjusted so that each core had one L of overlying water. Six cores per site (eighteen total) were wrapped to exclude light and incubated at room temperature (~22°C). The overlying water of each core was bubbled with air for 24 h, then half the cores per site continued to be bubbled with air (oxic conditions promoting aerobic processes), while the other half were bubbled with N2 (anoxic conditions promoting anaerobic processes). The cores were incubated for a month (from 22 July 2020 to 24 August 2020), and aluminum sulfate (Al2(SO4)3) was added to each core at a rate of 10:1 Al:P molar ratio, where the P was based on the site average of the SRP release rates estimated across all cores and conditions (see results for alum mass used). Three cores, one from each site, were incubated at room temperature on the windowsill with exposure to natural light.

Water samples (~50 mL) were removed from the overlying water of each core (including those exposed to natural light) at almost daily intervals for the first week and then every other day throughout the incubation. The sample was filtered (0.45 µm), acidified with HCl to pH<2, and then analyzed for SRP at the AWRC water quality lab. The overlying water in the cores was maintained at a volume of 1 L using filtered (0.45 µm) lake water with SRP near method detection limits (=0.005 mg L-1). The mass of SRP in the overlying water was estimated based on measured SRP concentrations and the mass balance of water collected and replaced.

Sediment P release rates (mg m-2 d-1) were calculated using the linear regression of SRP mass accumulation in water over eight days during the incubation (mg d-1) divided by the inside core area (0.005 m2). The SRP mass was corrected for the withdrawal and addition of SRP during sampling. In general, the first eight days were used because that represented the period of linear increase from the start of the incubation and following alum addition. An analysis of variance was used to evaluate whether sediment SRP release rates were different between sites and treatments (oxic versus anoxic), as well as before and after alum addition at sites within a treatment (P=0.05).

We inferred sediment equilibrium P concentrations (i.e., EPC0) from SRP concentration dynamics over time in the sediment-water cores; EPC0 is the concentration at which no net adsorption or desorption occurs from bottom sediments (Taylor and Kunishi, 1971; Froelich, 1988; Haggard et al., 1999) when overlying water is well oxygenated. The sediment EPC0 was estimated from the plateau SRP concentration in the overlying water of the cores incubated under oxic conditions. These SRP and EPC0 concentrations were useful in understanding nutrient availability over the course of the study at this lake.

Results

Sediment Incubations July 2020

Initial SRP concentrations in the overlying waters across all the collected sediment cores were low (=0.007 mg L-1; fig. 2), matching those typically observed in the lake water across the three monitoring sites. Two of the three cores sitting on the windowsill exposed to light showed increasing SRP concentrations through 6 d (up to 0.179 mg L-1) and then decreased below 0.007 mg L-1 through 12 d of the incubation; SRP remained low through the rest of the incubation in these cores. The SRP concentrations in the third core in the window were always less than the MDL (=0.005 mg L-1) throughout the incubation.

Figure 2. Mean soluble reactive phosphorus (SRP) concentrations in overlying water of cores incubated under oxic and anoxic conditions to maintain aerobic and anaerobic processes, respectively, from three sites at Lake Fayetteville (dam, inlet, and mid-lake) over time (A); mean (± standard deviation) SRP fluxes from sediment cores under oxic and anoxic conditions from three sites at Lake Fayetteville (dam, inlet, and mid-lake) prior to aluminum sulfate (alum) treatment (B), and SRP fluxes after alum treatment of sediment cores (C). [Letters in graphs B and C show differences in aerobic and anaerobic incubations, while numbers show differences in site means before and after alum.]

SRP concentrations in the overlying water of the cores incubated under oxic conditions (i.e., bubbled with air) slowly increased through the first 8 d. After 8 d, SRP concentration leveled off in some cores, continued to increase in others, and even slightly decreased in others (fig. 2). The minimum SRP concentration in the overlying water of the aerobic treatments was 0.018 mg L-1 from 8 to 23 d, and the maximum was 0.058 mg L-1. The average SRP was 0.028 mg L-1 in the overlying water across all cores, sites and between 8 to 23 d intervals. The range in average SRP from 8 to 23 d across the aerobic treatments was 0.022 to 0.038 mg L-1, likely marking dissolved P equilibrium (i.e., EPC0) between water and sediments.

The mass of SRP accumulating in the overlying water significantly (R2=0.88, P<0.01) increased over the first 8 d of the incubation across all aerobic cores. This resulted in SRP release rates that were not significantly different between the inlet, mid and dam sites at Lake Fayetteville (fig. 2). The SRP release rates ranged from 0.46 to 1.20 mg m-2 d-1 across all cores where the overlying water was bubbled with air (i.e., aerobic conditions), and the means were 0.88, 0.69, and 0.99 mg m-2 d-1 of the cores from the inlet, mid, and dam sites, respectively.

At 23 d, the overlying water of all cores was treated with alum; the doses ranged from 0.225, 0.450, and 0.900 g alum per core across the replicates at each site. The SRP concentrations in overlying water of all cores incubated under aerobic conditions dropped below the MDL after alum treatment (=0.003 mg L-1; fig. 2). The SRP release rates were significantly less under aerobic treatment following alum treatment, except at the mid lake site (fig. 2). The SRP release rates following alum treatment were near zero and not significantly different between sites, where release rates from individual cores ranged from -0.26 to 0.33 mg m-2 d-1.

On the other hand, SRP concentrations in the overlying water of the cores incubated under anoxic conditions (i.e., bubbled with N2) increased to 0.254 mg L-1 on average after 8 d, ranging from 0.130 to 0.398 mg L-1 across cores (fig. 2). SRP concentrations in the overlying water continued to increase in some cores, while concentrations in the overlying water of other cores leveled off and even decreased in the later part of the 23-d incubation. The maximum SRP concentration in the overlying water of the anaerobic cores was 0.727 mg L-1 between 8 and 23 d.

The mass of SRP accumulating in the overlying water significantly (R2=0.81, P<0.01) increased over time in the cores during the first 8 d of incubation under anoxic conditions. The rate of increase, or the slope of these relations, was much greater under anoxic conditions compared to oxic. The SRP release rates were not significantly different across the inlet, mid, or dam sites, averaging 7.61, 5.15, and 7.70 mg m-2 d-1, respectively (fig. 2). The SRP release rates ranged from 3.56 to 11.27 mg m-2 d-1 across all cores under anaerobic conditions, where the minimum and maximum were both from cores collected at the dam site.

At 23 d, alum treatment at the same doses used in aerobic cores was effective at reducing SRP in the overlying water of the cores incubating under anaerobic conditions, almost immediately knocking SRP down to 0.061 mg L-1 or less. The next day after alum treatment the SRP concentration in the overlying waters was less than the MDL on average (0.004 mg L-1) across all cores, and the maximum concentration was more than an order of magnitude less (0.010 mg L-1) than before alum dosing. SRP concentration in some anaerobic cores started slowly increasing, but the rate of increase was not significantly different from zero (P=0.07). Alum treatment significantly reduced SRP release rates across all sites under anoxic conditions (fig. 2).

Lake Water Quality March to September 2020

The dissolved nutrient supply in the water samples from Lake Fayetteville follows an annual pattern (see also Grantz et al., 2014), which likely has an influence on cyanobacterial HABs and microcystin production (fig. 3). Dissolved NH4-N is generally low (<0.1 mg L-1), fluctuating with intermittent peaks in concentrations throughout spring to fall 2020 across all three sites (fig. 3B). Dissolved NN started out greater than 1 mg L-1 in spring and sharply reduced over time up to June 2020, when NN was less than 0.05 mg L-1 except for episodic peaks near the inlet likely related to watershed rainfall-runoff (fig. 4). The Soluble reactive P supply is greatest in spring (max=0.014 mg L-1), declining to concentrations of 0.005 mg L-1 or less and persisting through summer (fig. 3C). The dissolved nutrient supply and molar N:P ratio initially favors P limited conditions for algal growth through spring, but the limited dissolved nutrient supply during the summer months suggests strong co-limitation of algal growth.

Figure 3. Microcystin, dissolved, and total nutrients, and raw fluorescence (RFU) of chlorophyll (CHL) and phycocyanin (PC), and PC:CHL ratio of water samples collected from March through September of 2020 at three sampling sites at Lake Fayetteville, Arkansas. [Note: NN and SRP are open symbols on total nutrient graphs.]

Total nutrient concentrations in the water samples from Lake Fayetteville followed a pattern similar to the dissolved nutrient supply but were an order of magnitude greater (figs. 3A and C). Total N concentrations started out with NN making up the majority of TN at the beginning of spring (~80%), and then as NN decreased up to June 2020, TN concentrations fluctuated between 1.20 and 2.00 mg L-1, averaging 1.57 mg L-1 (fig. 3A). After NN was hardly measurable, TN concentrations ranged from 0.72 to 1.61 mg L-1, averaging just greater than 1 mg L-1 from June through September 2020. Phosphorus, on the other hand, was more variable between sites through June 2020 compared to after and through September. Overall, TP concentrations averaged 0.50 mg L-1, ranging from 0.017 to 0.128 mg L-1 during this study (fig. 3C).

Figure 4. Thresholds and hierarchical structure in splits with total microcystin concentrations at Lake Fayetteville using only nutrient concentrations and algal pigment parameters at all three sites from March to September 2020, Lake Fayetteville, Arkansas.

The surrogates for algal and cyanobacterial biomass showed some interesting patterns from March through September 2020 (figs. 3D and E). Phycocyanin RFUs started relatively low in March (=2000 RFUs) and peaked in late April 2020 (31465 RFU). Phycocyanin RFUs were then variable into early summer and then slowly increased through summer and into fall, except for a sharp decrease in late August. Chlorophyll RFUs were more variable through the study period, ranging from 191 to 14730 RFU with an average of ~4800 RFU. The ratio of these RFUs (PC:CHL) started out less than one in March (fig. 3D), suggesting that chlorophyll is coming from a mix of algae and cyanobacteria. In mid-April, this ratio increased and averaged 3.9 through mid-June with a maximum of 8.9; this suggests that total chlorophyll was coming almost solely from cyanobacteria. In summer 2020, PC:CHL was ~1 and more stable through the growing season. This ratio slightly increased until a decrease in late August and then a sharp decline in late September (PC:CHL=0.56).

Measured Chl-a pigments more closely followed the pattern in phycocyanin RFUs (r=0.81, P<0.01) than CHL RFUs (r=0.17, P=0.11). The combination of PC and CHL RFUs explained 68 percent of the variability in measured pigments (Chl-a = 0.00101·CHL + 0.00263·PC + 6.85, P<0.01), which was not much stronger than PC alone. These findings suggest that cyanobacteria produced the most Chl-a at Lake Fayetteville during the growing season when compared to other algal groups. 

The least Chl-a concentrations were measured in early March 2020 (4.85 µg L-1, on average), and Chl-a pigments sharply increased into late April (90.6 µg L-1) and then decreased through the growing season (fig. 3E). The average Chl-a concentration during the growing season (i.e., May through September) was 33.9 µg L-1, ranging from 11.3 to 85.7 µg L-1 across the three sites; this reflected the hypereutrophic nature of this small impoundment. Chlorophyll-a concentrations were correlated with both TN (r=0.39, P<0.01) and TP (r=0.44, P<0.01) across all water samples collected from these three sites at Lake Fayetteville.

Microcystin concentrations were consistent across the three sampling sites and followed a pattern that inversely mirrored that of dissolved nutrients but mirrored surrogates for algal and cyanobacterial biomass (fig. 3F). These observations included the following:

  1. Microcystin was greatest when the dissolved nutrient supply was low and hardly measurable (especially NN, which was 0.014 mg L-1 on average and < 0.064 mg L-1);
  2. Microcystin was greatest when PC:CHL RFU averaged 1.3, which suggests most of the total chlorophyll was from cyanobacteria, but not at peak community dominance; and
  3. Microcystin was less than the reporting limit (0.3 µg L-1) when measured, Chl-a concentrations were greatest.

Microcystin concentrations were elevated from mid-June to mid-August in water samples collected just below the surface from all three sites at Lake Fayetteville, averaging 2.51 µg L-1 with a 4.75 µg L-1 max.

Microcystin Thresholds and Hierarchy March to September 2020

At Lake Fayetteville, microcystin showed significant change points (CP) or thresholds with most physicochemical, nutrient, and algal parameters measured in the water samples during the routine monitoring (table 1; nCPA, Pperm=0.05). The basic physicochemical properties showed thresholds, including water temperature, pH, and conductivity. Water temperature (CP, 27°C) explained 44% of the variation in microcystin concentrations, where mean microcystin concentration was almost ten times greater when the water temperatures exceeded the CP (1.930 µg L-1) compared with that when water temperature was less than the CP (0.159 µg L-1). The conductivity threshold was ~159 µS cm-1, and mean microcystin concentration was greater when conductivity was less than this CP. Greater mean microcystin concentration (1.241 µg L-1) occurred above the pH CP of 8.5 compared to when pH was less than 8.5 (0.132 µg L-1). Dissolved oxygen and rainfall within 24 hours to one week did not show significant CPs (P>0.05) with total microcystin.

Table 1. Non-parametric changepoints (CP) with 5% to 95% confidence intervals (CI) for relationships between nutrient, algal, and physicochemical parameters and microcystin concentrations at Lake Fayetteville; CPs and CIs shown only when statistically significant (Pperm=0.050).
ParameterCP (5% - 95% CI)ppermr2Mean Microcystin
< CP= CP
TN (mg/L)1.155 (1.075-1.380)0.0020.251.5660.233
DIN0.127 (0.009-0.176)0.0010.301.5300.0532
NN (mg/L)0.069 (0.025-0.093)0.0010.311.5600.0527
NH4-N (mg/L)0.005 (0.004 – 0.012)0.0190.111.5100.596
TP (mg/L)0.0605 (0.0365-0.062)0.0420.101.1540.0575
SRP (mg/L)-0.052---
Supply N:P115 (52-119)0.0010.271.6370.175
Particulate N:P42 (36-48)0.0010.190.1701.388
Chl a (µg/L)23.4 (23.1-29.3)0.0040.150.2411.328
Pheophytin (µg/L)-0.200---
Phycocyanin (RFU)3946 (3154-11726)0.0400.110.2091.221
CHL (RFU)4172 (3570-7726)0.0010.280.2551.667
PC:CHL1.4 (1.1-2.3)0.0050.181.5020.391
Water Temperature (°C)27.0 (25.8-28.5)0.0010.440.1591.930
pH8.53 (8.47-8.59)0.0160.140.1011.241
DO (mg/L)-0.356---
Conductivity (µS/cm)159 (137-162)0.0060.151.3160.228
24-hour precipitation (in)-0.404---
72-hour precipitation (in)-0.055---
1-week precipitation (in)-0.092---

Microcystin showed threshold responses to increasing nutrient concentrations and ratios both in the dissolved and particulate forms (table 1). The nutrient parameter that explained the most variation in microcystin with a single CP was NN (R2=0.31), where mean microcystin concentration (1.560 µg L-1) was over 13 times greater when NN was less than 0.069 mg L-1 compared to that above the threshold (0.053 µg L-1). Microcystin also showed a threshold response to DIN (0.127 mg L-1), which followed the same pattern as NN. Even NH4-N showed a significant CP with total microcystin, albeit the threshold (0.005 mg L-1) was below lab method detection limits. The mean microcystin concentrations were greater when TN concentrations were less than 1.155 mg L-1 and the proportion of DIN to TN was less than 0.13. Microcystin showed a significant CP with TP (0.061 mg L-1), with an order of magnitude greater mean microcystin concentration when TP was less than the CP (1.154 µg L-1), compared to greater than (0.058 µg L-1); SRP did not show a significant threshold (P>0.050). Microcystin showed significant CPs with the supply (115) and particulate (42) N:P ratios, where greater mean microcystin concentrations (1.4 to 1.6 µg L-1) were observed when ratios were greater than the thresholds. When we used CART with microcystin and only nutrients, a hierarchical structure existed with nutrients, but estimated secondary and tertiary splits were less than or in range with method detection limits (data not shown).

The four algal parameters that were measured in this study were connected to microcystin, and each displayed significant CPs (table 1). These included:

  1. a threshold with raw fluorescence of CHL (4172 RFUs) and PC (3946 RFUs);
  2. a threshold with the ratio PC:CHL at 1.4; and
  3. a threshold with measured Chl-a pigments (23.4 µg L-1).

The mean microcystin concentration was greater when RFUs and Chl-a concentrations were greater than those thresholds, but it was greater when the PC:CHL ratio was less than its CP.

When we used CART with microcystin, nutrients, and algal parameters, the hierarchical structure explained 82% of the variation in microcystin concentrations (fig. 4); the greatest mean microcystin concentration (3.9 µg L-1) occurred when NN was less than 0.069 mg L-1, PC:CHL was less than 1.5, CHL RFUs were greater than 4555, and TP was greater than 0.046 mg L-1. If we allowed the strongest individual CP (i.e., water temperature) to drive the hierarchical structure, that model explained 76% of the variability in microcystin and included secondary splits with CHL, and particulate N:P molar ratio (fig. 5). Within this structure, the greatest mean microcystin (3.5 µg L-1) occurred when the water temperature was greater than 27°C, CHL was greater than 3570 RFUs, and the particulate N:P ratio was less than 50.

These trees were pruned, which explains a bit of difference in deviation.

Discussion

Microcystin concentrations in US lakes are correlated with trophic status, and the relations between microcystin and environmental variables across and within lakes are complex and likely not linear (Graham et al., 2004). This broader observation was true at Lake Fayetteville in Arkansas, where microcystin showed thresholds and hierarchical structures with several physicochemical properties in the lake water. The strongest microcystin threshold was with water temperature (27°C) at Lake Fayetteville, which fits with observations that increasing water temperature promotes cyanobacteria growth (Yang et al., 2020) and increases total microcystin (Brutemark et al., 2015). However, water temperatures where microcystin was greatest at Lake Fayetteville were greater than the optimal range for cellular toxin production in some laboratory studies (20°C to 25°C; Walls et al., 2018). Some evidence suggests that increased temperatures promote cyanobacteria growth in lake water, but cellular toxin production (i.e., cell quota) is less at cooler temperatures (Peng et al., 2018). Whether it is due to increased cyanobacteria biomass or an increased toxin cell quota, when near-surface water temperatures exceeded 27°C at Lake Fayetteville, total microcystin concentrations (1.930 µg L-1) were almost ten times greater on average than at cooler temperatures (0.203 µg L-1). The effects of climate change and increasing temperatures may increase the occurrence and toxicity of HABs in lakes like Lake Fayetteville and globally, requiring adaptive mitigation strategies to reduce toxicity and improve water quality (Paerl and Barnard, 2020).

Figure 5. Thresholds and hierarchical structure in splits with total microcystin concentrations at Lake Fayetteville, Arkansas, using water temperature and all measured physicochemical properties at all three sites from 2020 March to 2020 September.

The effects of increased temperatures and climate change on HABs may be exacerbated by increasing nutrient inputs both internal and external to the lake. In fact, water temperature and dissolved nutrient supply are linked across lakes, where dissolved nutrient (particularly NO3 and SRP) supply in lake waters is greatest during cooler temperatures (i.e., winter). Then, as the cyanobacteria and eukaryotic algae uptake nutrients and grow with increasing water temperatures into summer, the dissolved nutrient supply (both NN and SRP) often decreases to below analytical detection limits. This is exactly what happens in Lake Fayetteville, and microcystin concentrations showed the next strongest threshold with N concentrations (0.069 mg L-1) in the lake water. At Lake Fayetteville in 2020, it was the beginning of June when both water temperatures approached 27°C and NN concentrations decreased to less than 0.1 mg L-1 in the lake water. Total microcystin concentrations in Lake Fayetteville increased after this point in time, persisting at 1 µg L-1 or above from mid-June through mid-August.

The SRP supply was also low during the 2020 growing season at Lake Fayetteville, especially compared to SRP concentrations of 0.085 mg L-1 in the lake water following the previous lake mixing in 2019. The supply of SRP may be low in the lake water below the surface, but bottom sediments are leaking SRP into the overlying water under both aerobic and anaerobic conditions. The hypolimnion has a buildup of SRP and NH4-N from bottom sediments during summer (Grantz et al., 2014), which diffuses slowly into the photic zone or is entrained during episodic breakdowns of the thermocline and especially following fall mixing when the lake is isothermal. This nutrient source potentially favors cyanobacteria, which have the ability to regulate buoyancy and can migrate lower in the water column to take advantage of these nutrients and then move back up again to optimal photic conditions (Reynolds et al., 1987). The chemistry of the lake water definitely suggests co-limitation of cyanobacteria by N and P, but the lake is an open system, not a closed one, where SRP release from bottom sediments must be considered as a potential (almost continuous) nutrient supply.

Lake sediments near the shore would most likely have enough dissolved oxygen in the overlying water to limit reductive dissolution, but they would also most likely try to keep dissolved P concentrations near their EPC0 (Taylor and Kunishi, 1971; Froleich, 1988; Haggard et al., 1999). Based on the core experiments, sediment EPC0 at Lake Fayetteville could be estimated to be ~0.03 mg L-1 (fig. 2), marking the concentration at which SRP stabilized in aerobic cores. Yet, we do not observe concentrations near this supply in the overlying water at Lake Fayetteville at our three near-shore sampling sites throughout the 2020 growing season. It is common in lakes to have SRP concentrations at analytical detection limits (<0.005 mg L-1), and this is likely because cyanobacteria and algae take the SRP up as fast as or faster than the sediments can release it into the overlying water (McCarty, 2019; Austin et al., 2020). We think these sediment P sources are fueling the cyanobacterial growth at this lake, and it might be these internal P sources that sustain the HABs. There are many options for lake managers to chemically inactivate this internal P source (e.g., see Welch and Cooke, 1999; Yin et al., 2021) and even reduce sediment EPC0 (Haggard et al., 2004), and these could be effective at Lake Fayetteville (fig. 2).

Microcystin thresholds were observed with nutrients and nutrient ratios at Lake Fayetteville, although some did not fit conventional thoughts about nutrients driving HABs. The observations at Lake Fayetteville showed that greater MC concentrations were observed at lower total and dissolved nutrient concentrations and supplies, whereas, high MC is generally thought of as occurring at greater nutrient concentrations across lakes (Paerl et al., 2001; Yuan and Pollard, 2017), and the probability of exceeding 1 µg MC L-1 increases with increasing nutrient and Chl-a pigment concentrations (Yuan et al., 2014). The within-lake difference is likely because of the NN and TN dynamics at Lake Fayetteville, where N is greatest early in the growing season (i.e., over winter and early spring) and NN makes up a large part of TN in this eutrophic reservoir. However, it does appear that MC concentrations in lakes are influenced by latitude and longitude and that globally, TN is a strong predictor of MC while TP is not (Buley et al., 2021). While TN might be important in predicting MC concentrations, TP may still play a significant role in predicting cyanobacterial biomass (Shan et al., 2020).

Microcystin has thresholds with N:P ratios, where MC decreases with increasing TN:TP ratios in lakes based on a global analysis (Harris et al., 2014). In Canadian lakes, MC was statistically greatest at intermediate molar TN:TP ratios (27 to 51) at high nutrients, and the probability of exceeding 1 µg L-1 was greatest at TN:TP molar ratios less than 80 in lakes with TN concentrations greater than a threshold of 2.6 mg L-1 (Scott et al., 2013). Lake Fayetteville generally fits these nutrient ratio patterns because MC was greatest when particulate molar PN:PP ratios exceeded 40, but were less than 63. However, we have to keep in mind the NN and TN dynamics that occur within individual reservoirs.

The measures of algal pigments and fluorescence at Lake Fayetteville do fit with the mental model that increased algal biomass coincides with elevated MC in lakes (e.g., see Yuan et al., 2014; Buley et al., 2021; Chaffin et al., 2021). Mean MC concentrations increased above the thresholds for CHL (4172 RFU), PC (3946 RFU), and Chl-a (23.4 µg L-1) for Lake Fayetteville, while mean MC was greatest when PC:CHL was less than 1.44; the mean MC concentrations above these thresholds were all greater than 1 µg L-1. The Chl-a threshold (and total nutrient concentrations) with total MC fits nicely with trophic status guidelines for lakes (i.e., ~25 µg L-1) (Nurmberg, 1996), suggesting cyanobacterial HABs and toxins are likely to occur under this trophic condition in Lake Fayetteville.

When we bring nutrients and algal parameters together, the hierarchical structure in our CART models tells an interesting story for Lake Fayetteville (fig. 4). When NN was present in the lake water (i.e., >0.069 mg L-1), total MC concentrations were less than method detection limits. When NN was low (=0.069 mg L-1), then the ratio of PC:CHL (RFUs) was important in the decision tree; this parameter is used to suggest whether the algal community is dominated by cyanobacteria or not (Thomson-Laing et al., 2020). A ratio exceeding 1.5 would suggest cyanobacteria were dominating, and when PC:CHL=1.5, the concentration of NH4-N played a role in toxin concentrations; if NH4-N exceeded 0.023 mg L-1, then mean total MC was 1.9 µg L-1. This could show a potential influence of NH4-N on MC production by cyanobacteria (e.g., see Chaffin et al., 2018) or the potential for cyanobacteria growth with intracellular toxins when this energetically favorable N supply is available (Herrero et al., 2001) at Lake Fayetteville. The influence of N supply forms (NH4-N, NN, and urea) on cyanobacteria and toxin production has been a recent research focus, even using lake water from Lake Fayetteville (Wagner et al., 2021).

While previous work has suggested that TP is not predictive of MC concentrations or production, it is predictive of cyanobacterial biomass and growth (Shan et al., 2020). At Lake Fayetteville, TP concentrations in the lake water played a significant role in our regression trees, especially at greater CHL RFUs (>4555) and total MC concentrations. Total MC concentrations were greatest (3.9 µg L-1) when biomass was likely elevated, and TP exceeded 0.046 mg L-1. At Lake Fayetteville, TP plays a role in total MC concentrations, but it is possible that it is just related to increased cyanobacterial biomass.

These physicochemical and pigment fluorescence thresholds with total microcystin concentration lead to a series of questions: are these predictors just increasing cyanobacterial growth with intracellular toxins or stimulating toxin production on a cellular basis within this hierarchical structure? Are these physicochemical and pigment thresholds and structures consistent across years at Lake Fayetteville? Lake Fayetteville has recurring cyanobacterial HABs, and nutrient dynamics and temporal patterns in 2020 were like other years with intensive temporal monitoring (Grantz et al., 2014), including historical data (Meyer, 1971). Cyanobacteria have likely been the dominant algal group at Lake Fayetteville for 50 years (Meyer, 1971), and, if we had historic data, we might have observed cyanotoxins even 50 years ago.

Conclusions

Lake Fayetteville provides a great opportunity to study the occurrence of cyanobacterial HABs and triggers resulting in toxin concentrations and production. The overlying water in Lake Fayetteville did not have much SRP available, despite the observations that sediment EPC0 suggested P release from lake sediments and incubation showed high release rates; sediments could be fueling the cyanobacterial blooms. Many of the physicochemical conditions measured at Lake Fayetteville showed significant thresholds with total microcystin concentrations, and hierarchical structure existed with nutrients and pigment raw fluorescence, explaining a large part of the variance in microcystin concentrations.

Lake Fayetteville has annual cyanobacterial HABs that produce toxins and nutrients (both N and P) from external and internal sources that are important drivers in the blooms and toxicity. The City of Fayetteville could focus on identifying external nutrient sources, using water quality and watershed characteristics (McCarty et al., 2018) in the relatively small 24 km2 watershed to reduce nutrient input. The lake could also be managed using chemical treatments to reduce internal nutrient availability and reduce sediment SRP release and EPC0.

Acknowledgments

Funding for this research was provided in part by the AWRC through the USGS WRRI 104B Base Funding Program, the USDA NIFA Hatch Program Project 2660, and the University of Arkansas System Division of Agriculture. The ability to use these funds to tackle specific issues like nutrient and environmental drivers of cyanobacterial HABs was key to this project, the published manuscript, and continued monitoring.

References

Anderson, D. M., Glibert, P. M., & Burkholder, J. M. (2002). Harmful algal blooms and eutrophication: Nutrient sources, composition, and consequences. Estuaries, 25(4), 704-726. https://doi.org/10.1007/BF02804901

Austin, B. J., Eagle, V., Evans-White, M. A., Scott, J. T., & Haggard, B. E. (2020). Sediment phosphorus release sustains nuisance periphyton growth when nitrogen is not limiting. J. Limnol., 79(3). https://doi.org/10.4081/jlimnol.2020.1913

Brooks, B. W., Lazorchak, J. M., Howard, M. D., Johnson, M.-V. V., Morton, S. L., Perkins, D. A.,... Steevens, J. A. (2016). Are harmful algal blooms becoming the greatest inland water quality threat to public health and aquatic ecosystems? Environ. Toxicol. Chem., 35(1), 6-13. https://doi.org/10.1002/etc.3220

Brutemark, A., J. Engstrom-Ost, A. Vehmaa, and E. Gorokhova. 2015. Growth, toxicity and oxidative stress of a cultured cyanobacterium (Dolichospermum sp.) under different Co2/pH and temperature conditions. Phycological Research 63:56-63.

Buley, R.P., H.E. Correia, A. Abebe, T.B. Issa, and A.E. Wilson. 2021. Predicting microcystin occurrence in freshwater lakes and reservoirs: assessing environmental variables. Inland Waters. 11:430-444.

Chaffin, J. D., Bratton, J. F., Verhamme, E. M., Bair, H. B., Beecher, A. A., Binding, C. E., ... & Zhou, X. (2021). The Lake Erie HABs Grab: A binational collaboration to characterize the western basin cyanobacterial harmful algal blooms at an unprecedented high-resolution spatial scale. Harmful algae, 108, 102080.

Chaffin, J. D., Davis, T. W., Smith, D. J., Baer, M. M., & Dick, G. J. (2018). Interactions between nitrogen form, loading rate, and light intensity on Microcystis and Planktothrix growth and microcystin production. Harmful Algae, 73, 84-97. https://doi.org/10.1016/j.hal.2018.02.001

City of Fayetteville. (2021). Lake Fayetteville. Retrieved from https://www.fayetteville-ar.gov/3130/Lake-Fayetteville

De’ath, G., & Fabricius, K. E. (2000). Classification and regression trees: A powerful yet simple technique for ecological data analysis. Ecology, 81(11), 3178-3192. https://doi.org/10.1890/0012-9658(2000)081[3178:CARTAP]2.0.CO;2

Ding, S., Chen, M., Gong, M., Fan, X., Qin, B., Xu, H.,... Zhang, C. (2018). Internal phosphorus loading from sediments causes seasonal nitrogen limitation for harmful algal blooms. Sci. Total Environ., 625, 872-884. https://doi.org/10.1016/j.scitotenv.2017.12.348

Environmental Protection Agency, (EPA). (2019). Recommended human health recreational ambient water quality criteria or swimming advisories for microcystins and cylindrospermopsin. EPA8222-F-19-001.

Environmental Working Group, (EWG). (2020). News reports of algae blooms, 2010 to present. Retrieved from https://www.ewg.org/interactive-maps/algal_blooms/map/

Froelich, P. N. (1988). Kinetic control of dissolved phosphate in natural rivers and estuaries: A primer on the phosphate buffer mechanism. Limnol. Oceanogr., 33, 649-668. https://doi.org/10.4319/lo.1988.33.4part2.0649

Graham, J.L., J.R. Jones, S.B. jones, J.A. Downing, and T.E. Clevenger. 2004. Environmental factors influencing microcystin distribution and concentrations in the Midwestern United States. Water Research 38:4395-4404.

Grantz, E.M., B.E. Haggard, and J.T. Scott. 2014. Stoichiometric imbalance in rates of nitrogen and phosphorus retention, storage, and recycling can perpetuate nitrogen deficiency in highly-productive reservoirs. Limnology and Oceanography 59:2203-2216.

Haggard, B. E., Ekka, S. A., Matlock, M. D., & Chaubey, I. (2004). Phosphate equilibrium between stream sediments and water: Potential effect of chemical amendments. Trans. ASAE, 47(4), 1113-1118. https://doi.org/10.13031/2013.16584

Haggard, B. E., Stanley, E. H., & Hyler, R. (1999). Sediment-phosphorus relationships in three Northcentral Oklahoma streams. Trans. ASAE, 42(6), 1709-1714. https://doi.org/10.13031/2013.13333

Harris, T. D., Wilhelm, F. M., Graham, J. L., & Loftin, K. A. (2014). Experimental manipulation of TN:TP ratios suppress cyanobacterial biovolume and microcystin concentration in large-scale in situ mesocosms. Lake Reserv. Manag., 30(1), 72-83. https://doi.org/10.1080/10402381.2013.876131

Herrero, A., Muro-Pastor, A. M., & Flores, E. (2001). Nitrogen control in cyanobacteria. J. Bacteriol., 183(2), 411-425. https://doi.org/10.1128/JB.183.2.411-425.2001

Jankowiak, J., Hattenrath-Lehmann, T., Kramer, B. J., Ladds, M., & Gobler, C. J. (2019). Deciphering the effects of nitrogen, phosphorus, and temperature on cyanobacterial bloom intensification, diversity, and toxicity in western Lake Erie. Limnol. Oceanogr., 64(3), 1347-1370. https://doi.org/10.1002/lno.11120

Kane, D. D., Conroy, J. D., Peter Richards, R., Baker, D. B., & Culver, D. A. (2014). Re-eutrophication of Lake Erie: Correlations between tributary nutrient loads and phytoplankton biomass. J. Great Lakes Res., 40(3), 496-501. https://doi.org/10.1016/j.jglr.2014.04.004

King, R. S., & Richardson, C. J. (2003). Integrating bioassessment and ecological risk assessment: An approach to developing numerical water-quality criteria. Environ. Manag., 31(6), 795-809. https://doi.org/10.1007/s00267-002-0036-4

Lapointe, B. E., Herren, L. W., & Paule, A. L. (2017). Septic systems contribute to nutrient pollution and harmful algal blooms in the St. Lucie Estuary, Southeast Florida, USA. Harmful Algae, 70, 1-22. https://doi.org/10.1016/j.hal.2017.09.005

Lapointe, B. E., Herren, L. W., Debortoli, D. D., & Vogel, M. A. (2015). Evidence of sewage-driven eutrophication and harmful algal blooms in Florida’s Indian River Lagoon. Harmful Algae, 43, 82-102. https://doi.org/10.1016/j.hal.2015.01.004

McCarty, J. A. (2019). Algal demand drives sediment phosphorus release in a shallow eutrophic cove. Trans. ASABE, 62(5), 1315-1324. https://doi.org/10.13031/trans.13300

McCarty, J. A., Matlock, M. D., Scott, J. T., & Haggard, B. E. (2018). Risk indicators for identifying critical source areas in five Arkansas watersheds. Trans. ASABE, 61(3), 1025-1032. https://doi.org/10.13031/trans.12699

Meyer, R. L. (1971). A study of phytoplankton dynamics in Lake Fayetteville as a means of assessing water quality. PUB010A. 68. Arkansas Water Resource Center. Retrieved from https://scholarworks.uark.edu/awrctr/357

Nürnberg, G. K. (1996). Trophic state of clear and colored, soft- and hardwater lakes with special consideration of nutrients, anoxia, phytoplankton and fish. Lake Reserv. Manag., 12(4), 432-447. https://doi.org/10.1080/07438149609354283

Orihel, D. M., Schindler, D. W., Ballard, N. C., Graham, M. D., O’Connell, D. W., Wilson, L. R., & Vinebrooke, R. D. (2015). The “nutrient pump:” Iron-poor sediments fuel low nitrogen-to-phosphorus ratios and cyanobacterial blooms in polymictic lakes. Limnol. Oceanogr., 60(3), 856-871. https://doi.org/10.1002/lno.10076

Paerl, H.W., M.A. Barnard. 2020. Mitigating the global expansion of harmful cyanobacterial blooms: moving targets in a human- and climatically-altered world. Harmful Algae 96:101845

Paerl, H. W., Fulton, R. S., Moisander, P. H., & Dyble, J. (2001). Harmful freshwater algal blooms, with an emphasis on cyanobacteria. Scientific World J., 1, 139109. https://doi.org/10.1100/tsw.2001.16

Paerl, H. W., Gardner, W. S., Havens, K. E., Joyner, A. R., McCarthy, M. J., Newell, S. E.,... Scott, J. T. (2016a). Mitigating cyanobacterial harmful algal blooms in aquatic ecosystems impacted by climate change and anthropogenic nutrients. Harmful Algae, 54, 213-222. https://doi.org/10.1016/j.hal.2015.09.009

Paerl, H. W., Scott, J. T., McCarthy, M. J., Newell, S. E., Gardner, W. S., Havens, K. E.,... Wurtsbaugh, W. A. (2016b). It takes two to tango: When and where dual nutrient (N & P) reductions are needed to protect lakes and downstream ecosystems. Environ. Sci. Technol., 50(20), 10805-10813. https://doi.org/10.1021/acs.est.6b02575

Peng, G., R.M. Martin, S.P. Dearth, X. Sun, G.L. Boyer, S.R. Campagna, S. Lin, and S.W. Wilhelm. 2018. Seasonally relevant cool temperatures interact with N chemistry to increase microcystins produced in lab cultures of Microcystis aeruginosa NIES-843. Environmental Science and Technology 52:4127-4136.

Qian, S. S., King, R. S., & Richardson, C. J. (2003). Two statistical methods for the detection of environmental thresholds. Ecol. Model., 166(1), 87-97. https://doi.org/10.1016/S0304-3800(03)00097-8

Reynolds, C. S., Oliver, R. L., & Walsby, A. E. (1987). Cyanobacterial dominance: The role of buoyancy regulation in dynamic lake environments. N. Z. J. Mar. Freshwater Res., 21(3), 379-390. https://doi.org/10.1080/00288330.1987.9516234

Scott, J. T., McCarthy, M. J., Otten, T. G., Steffen, M. M., Baker, B. C., Grantz, E. M.,... Paerl, H. W. (2013). Comment: An alternative interpretation of the relationship between TN:TP and microcystins in Canadian lakes. Can. J. Fish. Aquat.Sci., 70(8), 1265-1268. https://doi.org/10.1139/cjfas-2012-0490

Shan, K., X. Wang, H. Yang, B. Zhou, L. song, and M. Shang. 2020. Use statistical machine learning to detect nutrient thresholds in Microcystis blooms and microcystin management. Harmful Algae 2020:101807.

Taylor, A. W., & Kunishi, H. M. (1971). Phosphate equilibria on stream sediment and soil in a watershed draining an agricultural region. J. Agric. Food. Chem., 19(5), 827-831. https://doi.org/10.1021/jf60177a061

Therneau, T., & Atkinson, B. (2019). rpart: Recursive Partitioning and Regression Trees. R Package version 4.1-15. Retrieved from https://CRAN.R-project.org/package=rpart

Thomson-Laing, G., Puddick, J., & Wood, S. A. (2020). Predicting cyanobacterial biovolumes from phycocyanin fluorescence using a handheld fluorometer in the field. Harmful Algae, 97, 101869. https://doi.org/10.1016/j.hal.2020.101869

Wagner, N. D., Quach, E., Buscho, S., Ricciardelli, A., Kannan, A., Naung, S. W.,... Scott, J. T. (2021). Nitrogen form, concentration, and micronutrient availability affect microcystin production in cyanobacterial blooms. Harmful Algae, 103, 102002. https://doi.org/10.1016/j.hal.2021.102002

Walls, J.T., K.H. Wyatt, J.C. Doll, E.M. Rubenstein, and A.R. Rober. 2018. Hot and toxic: Temperature regulates microcystin release from cyanobacteria. Science of The Total Environment 610–611: 786-795.

Wurtsbaugh, W. A., Paerl, H. W., & Dodds, W. K. (2019). Nutrients, eutrophication and harmful algal blooms along the freshwater to marine continuum. WIREs Water, 6(5), e1373. https://doi.org/10.1002/wat2.1373

Yang, Z., M. Zhang, and X. Shi. 2020. Temperature triggers the annual cycle of Microcystis, comparable results from the laboratory and a large shallow lake. Chemosphere 260:127543.

Yin, H., Yang, C., Yang, P., Kaksonen, A. H., & Douglas, G. B. (2021). Contrasting effects and mode of dredging and in situ adsorbent amendment for the control of sediment internal phosphorus loading in eutrophic lakes. Water Res., 189, 116644. https://doi.org/10.1016/j.watres.2020.116644

Yuan, L.L., and A.I. Pollard. 2017. Using national-scale data to develop nutrient-microcystin relationships that guide management decisions. Environmental Science and Technology 51:6972-6980.

Yaun, L.L., A.I. Pollard, S. Prather, J.L. Oliver, and L. D’Anglada. 2014. Managing microcystin: Identifying national-scale thresholds for total nitrogen and chlorophyll a. Freshwater biology 59:1970-1981.