Using National-Scale Data To Develop Nutrient-Microcystin Relationships That Guide Management Decisions

Metagenomic insights into cyanotoxin dynamics in a Mexican subtropical lake

Valle de Bravo is a vital water supply for part of the metropolitan area of the Valle de Mexico megacity, providing 30% of Mexico City's water demand. This water body has experienced an acceleration in its trophic status, going from oligotrophic to eutrophic in just a few years. This temperate lake (at a tropical latitude) is in a persistent bloom dominated by a variety of co-occurring cyanobacteria, many of which have toxigenic potential based on microscopic identification, that makes it difficult or even impractical to identify the cyanotoxin producers. To unravel this complexity and directly identify the toxigenic genera, we showed that integrating classical approaches with metagenomic is required. We first characterized, from genes to metagenomes assembled genomes, the toxigenic Cyanobacteria. We found that Microcystis was the most dominant cyanobacterial genus and the sole carrier of the mcy operon, making it the only microcystin producer. We then quantified twenty-one different cyanopeptides, including twelve microcystin congeners using a high-performance liquid chromatography-high-resolution. Nine microcystins (MCs) and the emerging cyanotoxin anabaenopeptin-A and -B were found at varying concentrations throughout the year, with MC-LA being the most common and abundant. Our findings, constrained by our sampling strategy, indicate that conventional cyanotoxin biomarkers (e.g., toxin mcy genes) were not consistently reliable indicators of cyanotoxin concentrations in this freshwater system. In this study, we followed the dynamics of the cyanobacterial community and the associated cyanopeptides with unprecedented resolution. Our results have implications for better management of toxic blooms in this freshwater system, which supplies drinking water to more than 7 million people in the megalopolis of Valle de México.

Transcriptional profiles of Microcystis reveal gene expression shifts that promote bloom persistence in in situ mesocosms

Harmful algal blooms caused by cyanobacteria threaten aquatic ecosystems, the economy, and human health. Previous work has tried to identify the mechanisms that allow blooms to form, focusing on the role of nutrients. However, little is known about how introduced nutrients influence gene expression in situ. To address this knowledge gap, we used in situ mesocosms initiated with water experiencing a Microcystis bloom. We added pulses of nutrients that are commonly associated with anthropogenic sources to the mesocosms for 72 hours and collected samples for metatranscriptomics to examine how the physiological function of Microcystis and bloom status changed. The addition of nitrogen (N) as urea, but not the addition of PO4, resulted in conspicuous bloom persistence for at least 9 days after the final introduction of nutrients. The addition of urea initially resulted in the upregulation of photosynthesis machinery, as well as phosphate, carbon, and N transport and metabolism. Once Microcystis presumably became N-replete, upregulation of amino acid metabolism, microcystin biosynthesis, and other processes associated with biomass generation occurred. These capacities coincided with the upregulation of toxin-antitoxin systems, CRISPR-cas genes, and transposases suggesting that phage defense and genome rearrangement are critical in bloom persistence. Overall, our results show the stepwise transcriptional response of a Microcystis bloom to the introduction of nutrients, specifically urea, as it is sustained in a natural setting. The transcriptomic shifts observed herein may serve as markers of the longevity of blooms while providing insight into why Microcystis blooms over other cyanobacteria.IMPORTANCEHarmful algal blooms represent a threat to human health and ecosystems. Understanding why blooms persist may help us develop warning indicators of bloom persistence and create novel mitigation strategies. Using mesocosm experiments initiated with water with an active bloom, we measured the stepwise transcription changes of the toxin-producing cyanobacterium Microcystis in response to the addition of nutrients that are important in causing blooms. We found that nitrogen (N), but not phosphorus, promoted bloom longevity. The initial introduction of N resulted in the upregulation of genes involved in photosynthesis and N import. At later times in the bloom, upregulation of genes involved in biomass generation, phage protection, genomic rearrangement, and toxin production was observed. Our results suggest that Microcystis first fulfills nutritional requirements before investing energy in pathways associated with growth and protection against competitors, which allowed bloom persistence more than a week after the final addition of nutrients.

Immobilization of Microcystin by the Hydrogel-Biochar Composite to Enhance Biodegradation during Drinking Water Treatment

Microcystin-LR (MC-LR), the most common algal toxin in freshwater, poses an escalating threat to safe drinking water. This study aims to develop an engineered biofiltration system for water treatment, employing a composite of poly(diallyldimethylammonium chloride)-biochar (PDDA-BC) as a filtration medium. The objective is to capture MC-LR selectively and quickly from water, enabling subsequent biodegradation of toxin by bacteria embedded on the composite. The results showed that PDDA-BC exhibited a high selectivity in adsorbing MC-LR, even in the presence of competing natural organic matter and anions. The adsorption kinetics of MC-LR was faster, and capacity was greater compared to traditional adsorbents, achieving a capture rate of 98% for MC-LR (200 μg/L) within minutes to tens of minutes. Notably, the efficient adsorption of MC-LR was also observed in natural lake waters, underscoring the substantial potential of PDDA-BC for immobilizing MC-LR during biofiltration. Density functional theory calculations revealed that the synergetic effects of electrostatic interaction and π-π stacking predominantly contribute to the adsorption selectivity of MC-LR. Furthermore, experimental results validated that the combination of PDDA-BC with MC-degrading bacteria offered a promising and effective approach to achieve a sustainable removal of MC-LR through an "adsorption-biodegradation" process.

Microcystin concentrations and congener composition in relation to environmental variables across 440 north-temperate and boreal lakes

Understanding the environmental conditions and taxa that promote the occurrence of cyanobacterial toxins is imperative for effective management of lake ecosystems. Herein, we modeled total microcystin presence and concentrations with a broad suite of environmental predictors and cyanobacteria community data collected across 440 Canadian lakes using standardized methods. We also conducted a focused analysis targeting 14 microcystin congeners across 190 lakes, to examine how abiotic and biotic factors influence their relative proportions. Microcystins were detected in 30 % of lakes, with the highest total concentrations occurring in the most eutrophic lakes located in ecozones of central Canada. The two most commonly detected congeners were MC-LR (61 % of lakes) and MC-LA (37 % of lakes), while 11 others were detected more sporadically across waterbodies. Congener diversity peaked in central Canada where cyanobacteria biomass was highest. Using a zero-altered hurdle model, the probability of detecting microcystin was best explained by increasing Microcystis biomass, Daphnia and cyclopoid biomass, soluble reactive phosphorus, pH and wind. Microcystin concentrations increased with the biomass of Microcystis and other less dominant cyanobacteria taxa, as well as total phosphorus, cyclopoid copepod biomass, dissolved inorganic carbon and water temperature. Collectively, these models accounted for 34 % and 70 % of the variability, respectively. Based on a multiple factor analysis of microcystin congeners, cyanobacteria community data, environmental and zooplankton data, we found that the relative abundance of most congeners varied according to trophic state and were related to a combination of cyanobacteria genera biomasses and environmental variables.

Differential effects of nitrate and ammonium on the growth of algae and microcystin production by nitrogen-fixing Nostoc sp. and non-nitrogen-fixing Microcystis aeruginosa

Cyanotoxins produced by cyanobacteria are a significant threat to human health. However, their responses to nitrogen (N) supplies could differ between N-fixing and non-N-fixing species, which has been poorly understood. This study aimed to compare the responses of the non-N-fixing Microcystis aeruginosa and N-fixing Nostoc sp. to varying concentrations of nitrate and ammonium. This comparison had been conducted by analyzing chlorophyll-a contents, maximum quantum efficiencies of photosystem II, microcystin production, and related gene expressions. Our findings revealed that nitrate substantially stimulated the growth of both M. aeruginosa and Nostoc sp. with biomass increase by 366.2 ± 56.5 and 93.0 ± 14.0%, respectively, at 16 mg-N/L. In contrast, high ammonium concentrations suppressed their growth. Furthermore, the intracellular concentration of microcystins produced by M. aeruginosa was higher under high nitrate. Extracellular microcystins showed an opposite trend to increases in nitrate and ammonium. Ammonium increases the production and releases microcystin from Nostoc sp. N metabolism genes showed a similar trend with toxin formation genes, which were up-regulated under the high N treatments. This study provides valuable insights into the impacts of N supplies on growths of N- and non-N-fixing cyanobacteria, as well as microcystin production, which helps to develop effective strategies for managing cyanobacterial blooms.

Understanding the effects of phosphorus on diatom richness in rivers and streams using taxon-environment relationships

Changes in phosphorus concentrations affect periphytic diatom composition in streams, yet we rarely observe strong relationships between diatom richness and phosphorus. In contrast, changes in conductivity are strongly associated with differences in both diatom composition and richness. We hypothesised that we could better understand the mechanisms that control the phosphorus-richness relationship by examining relationships between phosphorus and the occurrence of individual diatom taxa, comparing these with relationships between conductivity and taxon occurrence, and documenting how niche breadths of taxa affect richness patterns. We estimated relationships between phosphorus and taxon occurrence using DNA metabarcoding data of diatoms collected from 1,811 sites distributed across the conterminous U.S.A. and contrasted patterns in these relationships with those between conductivity and taxon occurrence. The distribution of taxon optima for phosphorus was bimodal, with most optima located at either the maximum or minimum observed phosphorus concentration. The distribution of taxon optima for conductivity was unimodal. Niche breadths of taxa for phosphorus and for conductivity both generally increased with optimum values. The distribution of conductivity optima gave rise to a prominent hump-shaped relationship between richness and conductivity. The relationship between richness and phosphorus was also slightly hump-shaped, but this relationship would not be expected from the bimodal distribution of optima. Instead, we determined that broad niche breadths caused the hump-shaped relationship between richness and phosphorus. Our results highlight the nuanced effects that increased P loadings exert on diatom assemblages in rivers and streams and identify reasons that weak relationships between taxon richness and increased phosphorus have been observed. These findings allow us to better describe how excess phosphorus and subsets of taxa and their niche breadths contribute to patterns of taxa richness in diatom assemblages, and to improve the tools used to manage phosphorus pollution.

Temporal dynamics of microcystins in two reservoirs with different trophic status during the early growth stage of cyanobacteria

Harmful cyanobacterial blooms are increasing in frequency and severity, which makes their toxic secondary metabolites of microcystins (MCs) have been widely studied, especially in their distribution and influence factors in different habitats. However, the distribution of MCs on the early growth stage of harmful cyanobacteria and its influence factors and risks are still largely unknown. Thus, in the present study, two reservoirs (Lutian Reservoir and Lake Haitang) with different trophic status in China have been studied weekly from March to May in 2018, when the cyanobacteria communities were just in the early growth stage, to investigate the variation of MCs concentration and the relationships between MCs and environmental parameters. During the investigation, Lutian Reservoir and Lake Haitang were found to be mesotrophic and light eutrophic, respectively. In Lutian Reservoir, the concentration of EMCs (extracellular MCs) was obviously higher than that of IMCs (intracellular MCs) with a mean value of 0.323 and 0.264 μg/L, respectively. Meanwhile, the concentration of EMCs also fluctuated more sharply than that of IMCs. Congeners of IMC-YR and EMC-LR were respectively dominant in total concentrations of IMCs and EMCs. Unsurprisingly, in Lake Haitang, the concentrations of IMC and EMC were both significantly higher than that in Lutian Reservoir with a mean concentration of 0.482 and 0.472 μg/L, respectively. Differently, the concentration of MC-YR was dominant in both IMCs and EMCs, followed by MC-LR. In correlation analysis, the IMCs were significantly and positively correlated with the density and biomass of phytoplankton phyla and potential MCs-producing cyanobacteria and the parameters of water temperature (WT), nutrients, and organic matters. Similar results were also observed for EMCs. While the different variations of MCs in the two reservoirs might be primarily caused by the differences in WT, nutrients (especially phosphorus), organic matters, and the composition of MCs-producing cyanobacteria. In addition, the coexistence of the dominant species of Pseudoanabaena sp., which can produce a taste-and-odor compound of 2-methylisoborneol (2-MIB), might have a significant impact on the concentration and toxicity of MCs. Our results suggested that the risks posed by MCs at the early growth stage of cyanobacteria should also deserve our attention, especially in mesotrophic water bodies.

Revealing Physiochemical Factors and Zooplankton Influencing Microcystis Bloom Toxicity in a Large-Shallow Lake Using Bayesian Machine Learning

Toxic cyanobacterial blooms have become a severe global hazard to human and environmental health. Most studies have focused on the relationships between cyanobacterial composition and cyanotoxins production. Yet, little is known about the environmental conditions influencing the hazard of cyanotoxins. Here, we analysed a unique 22 sites dataset comprising monthly observations of water quality, cyanobacterial genera, zooplankton assemblages, and microcystins (MCs) quota and concentrations in a large-shallow lake. Missing values of MCs were imputed using a non-negative latent factor (NLF) analysis, and the results achieved a promising accuracy. Furthermore, we used the Bayesian additive regression tree (BART) to quantify how Microcystis bloom toxicity responds to relevant physicochemical characteristics and zooplankton assemblages. As expected, the BART model achieved better performance in Microcystis biomass and MCs concentration predictions than some comparative models, including random forest and multiple linear regression. The importance analysis via BART illustrated that the shade index was overall the best predictor of MCs concentrations, implying the predominant effects of light limitations on the MCs content of Microcystis. Variables of greatest significance to the toxicity of Microcystis also included pH and dissolved inorganic nitrogen. However, total phosphorus was found to be a strong predictor of the biomass of total Microcystis and toxic M. aeruginosa. Together with the partial dependence plot, results revealed the positive correlations between protozoa and Microcystis biomass. In contrast, copepods biomass may regulate the MC quota and concentrations. Overall, our observations arouse universal demands for machine-learning strategies to represent nonlinear relationships between harmful algal blooms and environmental covariates.

Numerical simulation of microcystin distribution in Liangxi River, downstream of Taihu Lake

Microcystins (MCs), the algal toxins produced by cyanobacteria, raised a worldwide concern in recent decades. Limited monitoring stations for MCs make it hard to map the MC spatial distribution in certain areas. To tackle such problems, we selected Liangxi River as our research area and developed an integrated model to get spatial continuous MC data without too many sampling sites, which integrates a hydro-environment model and an artificial neural network algorithm (ANN). The ANN algorithm can estimate concentration MCs via environmental factors. In this paper, we selected chl-a, TN, TP, NO 2 - , NO 3 - , NH₃ -N, and PO 4 3 - as stressors. The ANN model we established showed good performances both in train (R²  = 0.8407) and test set (R²  = 0.7543). In the hydro-environment model, by inputting river geometry and model boundary data, the spatial continuous water quality data could be simulated. The water quality data returned from the hydro-environmental model were used as input variables of the well-trained ANN model; the continuous MC data were derived. To evaluate this model on geo-mapping the MC distribution in Liangxi River, we compared the performance of this model and spatial interpolation on the test set, it turns out the integrated model showed a better performance. © 2020 Water Environment Federation PRACTITIONER POINTS: The cost of microcystin (MC) detection is too high for routine monitoring. We integrated regression method and hydro-environment model to predict MCs. Results derived from spatial interpolation are not robust in unmonitored area. The new integration model can minimize the drawback of spatial interpolation.

Nitrogen inputs best predict farm field nitrate leaching in the Willamette Valley, Oregon

Nitrate leaching is an important yet difficult to manage contribution to groundwater and surface water contamination in agricultural areas. We examine 14 farm fields over a four year period (2014-2017) in the southern Willamette Valley, providing 53 sets of annual, field-level agricultural performance metrics related to nitrogen (N), including fertilizer inputs, crop harvest outputs, N use efficiency (NUE), nitrate-N leaching and surplus N. Crop-specific nitrate-N leaching varied widely from 10 kg N ha-1yr-1 in hazelnuts to >200 kg N ha-1yr-1 in peppermint. Averaging across all sites and years, most leaching occurred during fall (60%) and winter (32%). Overall NUE was 57%. We used a graphical approach to explore the relationships between N inputs, surplus, crop N harvest removal and NUE by crop type. The blueberry site had high inputs and surplus, peppermint had high inputs but also high crop N removal and NUE and thus lower surplus, and most wheat crops had high NUE and evidence of using soil N. Annual N surplus was not well correlated with leaching, and leaching varied more by crop type and inputs. Grass seed and hazelnuts, which are dominant crop types in the southern Willamette Valley, were intermediate in terms of NUE, leaching and surplus. Of all performance metrics, N input was most closely aligned with field-level crop N harvest and nitrate leaching, therefore optimizing N inputs may well inform local efforts to reduce groundwater nitrate contamination.

Risk prediction of microcystins based on water quality surrogates: A case study in a eutrophicated urban river network

Microcystins (MCs), the toxic by-products from harmful algal bloom (HAB), have caused world-wide concern due to their acute toxicity in freshwater ecosystems. Most studies on HAB have been conducted for shallow freshwater lakes, such as Taihu Lake in China. However, algal blooms in urban rivers located downstream of eutrophicated lakes are also a serious problem for local administrators. It is important for them to know the current and potential risk level of MCs. This environmental issue is rarely reported or discussed. Within this context, we monitored MC concentrations in the Binhu River Network (BRN) in the algal bloom season (Aug, Sep, and Oct) in 2019. To note if the MC concentrations were dangerous, we used 1.0 μg/L suggested by the World Health Organization as the standard value. The proportions of MC samples violating the standard value were 31.78% (Aug), 21.14% (Sep) and 30.77% (Oct). We also designed two statistical models to predict MC concentrations and the possibility to exceed the standard level based on 10 water quality surrogates: Artificial Neural Network (ANN) and Logistic Regression (LR) models. These two models were trained and validated by the monitoring dataset (n = 224). Both models had good performances during training and testing. Although the water quality varied diversely both in spatial and temporal scale, Cluster Analysis (CA) could detect similarities among the samples and separated them into 3 classes, with each class denoting different types of rivers based on the 10 water quality surrogates. Then the ANN and LR were applied as a function of chl-a in each class; by gradually increasing chl-a concentration, we detected chl-a thresholds in class 1, 2, 3 were 25.5, 224, and 109.5 μg/L, respectively, when MCs have a 50% possibility to exceed standard level. The threshold values provided important implications for MC management in the BRN.

Nitrogen form, concentration, and micronutrient availability affect microcystin production in cyanobacterial blooms

Harmful algal blooms (HABs) are increasing in magnitude, frequency, and duration caused by anthropogenic factors such as eutrophication and altered climatic regimes. While the concentrations and ratios of nitrogen (N) and phosphorus are correlated with bloom biomass and cyanotoxin production, there is less known about how N forms and micronutrients (MN) interact to regulate HABs and cyanotoxin production. Here, we used two separate approaches to examine how N and MN supply affects cyanobacteria biomass and cyanotoxin production. First, we used a Microcystis laboratory culture to examine how N and MN concentration and N form affected the biomass, particulate N, and microcystin-LR concentration and cell quotas. Then, we monitored the N, iron, molybdenum, and total microcystin concentrations from a hypereutrophic reservoir. From this hypereutrophic reservoir, we performed a community HAB bioassay to examine how N and MN addition affected the biomass, particulate N, and microcystin concentration. Microcystis laboratory cultures grown in high urea and MN conditions produced more biomass, particulate N, and had similar C:N stoichiometry, but lower microcystin-LR concentrations and cell quotas when compared to high nitrate and MN conditions. Our community HAB bioassay revealed no interactions between N concentration and MN addition caused by non-limiting MN background concentrations. Biomass, particulate N, and microcystin concentration increased with N addition. The community HAB amended with MN resulted in greater microcystin-LA concentration compared to non-MN amended community HABs. Our results highlight the complexity of how abiotic variables control biomass and cyanotoxin production in both laboratory cultures of Microcystis and community HABs.

Occurrence and risk assessment of microcystin and its relationship with environmental factors in lakes of the eastern plain ecoregion, China

The frequent occurrence of microcystins (MCs) in freshwater poses serious threats to the drinking water safety and health of human beings. Although MCs have been detected in individual fresh waters in China, little is known about their occurrence over a large geographic scale. An investigation of 30 subtropical lakes in eastern China was performed during summer 2018 to determine the MCs concentrations in water and their possible risk via direct water consumption to humans, and to assess the associated environmental factors. MCs were detected in 28 of 30 lakes, and the highest mean MCs concentrations occurred in Lake Chaohu (26.7 μg/L), followed by Lake Taihu (3.11 μg/L). MC-LR was the primary variant observed in our study, and MCs were mainly produced by Microcystis, Anabaena (Dolicospermum), and Oscillatoria in these lakes. Replete nitrogen and phosphorus concentrations, irradiance, and stable water column conditions were critical for dominance of MC-producing cyanobacteria and high MCs production in our study. Hazard quotients indicated that human health risk of MCs in most lakes was at moderate or low levels except Lakes Chaohu and Taihu. Nutrient control management is recommended to decrease the likelihood of high MCs production. Finally, we recommend the regional scale thresholds of total nitrogen and total phosphorus concentrations of 1.19 mg/L and 7.14 × 10^-2 mg/L, respectively, based on the drinking water guideline of MC-LR (1 μg/L) recommended by World Health Organization. These targets for nutrient control will aid water quality managers to reduce human health risks created by exposure to MCs.

Modeling hypolimnetic dissolved oxygen depletion using monitoring data

Eutrophication increases hypoxia in lakes and reservoirs, causing deleterious effects on biological communities. Quantitative models would help managers develop effective strategies to address hypoxia issues, but most existing models are limited in their applicability to lakes with temporally resolved dissolved oxygen data. We describe a hierarchical Bayesian model that predicts dissolved oxygen in lakes based on a mechanistic understanding of the factors that influence the development of hypoxia during summer stratification. These factors include the days elapsed since stratification, dissolved organic carbon concentration, lake depth, and chlorophyll concentration. We demonstrate that the model can be fit to two datasets: one in which temporally resolved dissolved oxygen profiles were collected from 20 lakes in a single state and one in which single profiles were collected from 381 lakes across the United States. Analyses of these two datasets yielded similar relationships between volumetric oxygen demand and chlorophyll concentration, suggesting that the model structure appropriately represented the effects of eutrophication on oxygen depletion. Combining both datasets in a single model further improved the precision of predictions.

Use statistical machine learning to detect nutrient thresholds in Microcystis blooms and microcystin management

The frequency of toxin-producing cyanobacterial blooms has increased in recent decades due to nutrient enrichment and climate change. Because Microcystis blooms are related to different environmental conditions, identifying potential nutrient control targets can facilitate water quality managers to reduce the likelihood of microcystins (MCs) risk. However, complex biotic interactions and field data limitations have constrained our understanding of the nutrient-microcystin relationship. This study develops a Bayesian modelling framework with intracellular and extracellular MCs that characterize the relationships between different environmental and biological factors. This model was fit to the across-lake dataset including three bloom-plagued lakes in China and estimated the putative thresholds of total nitrogen (TN) and total phosphorus (TP). The lake-specific nutrient thresholds were estimated using Bayesian updating process. Our results suggested dual N and P reduction in controlling cyanotoxin risks. The total Microcystis biomass can be substantially suppressed by achieving the putative thresholds of TP (0.10 mg/L) in Lakes Taihu and Chaohu, but a stricter TP target (0.05 mg/L) in Dianchi Lake. To maintain MCs concentrations below 1.0 μg/L, the estimated TN threshold in three lakes was 1.8 mg/L, but the effect can be counteracted by the increase of temperature. Overall, the present approach provides an efficient way to integrate empirical knowledge into the data-driven model and is helpful for the management of water resources.

The implications of Simpson's paradox for cross-scale inference among lakes

Using cross-sectional data for making ecological inference started as a practical means of pooling data to enable meaningful empirical model development. For example, limnologists routinely use sample averages from numerous individual lakes to examine patterns across lakes. The basic assumption behind the use of cross-lake data is often that responses within and across lakes are identical. As data from multiple study units across a wide spatiotemporal scale are increasingly accessible for researchers, an assessment of this assumption is now feasible. In this study, we demonstrate that this assumption is usually unjustified, due largely to a statistical phenomenon known as the Simpson's paradox. Through comparisons of a commonly used empirical model of the effect of nutrients on algal growth developed using several data sets, we discuss the cognitive importance of distinguishing factors affecting lake eutrophication operating at different spatial and temporal scales. Our study proposes the use of the Bayesian hierarchical modeling approach to properly structure the data analysis when data from multiple lakes are employed.

Combining national and state data improves predictions of microcystin concentration

Data collected from lakes at national (regional) scales and state (local) scales can provide different insights regarding relationships between environmental factors and biological responses, and combining these two types of data can potentially yield more precise and accurate understanding of ecological phenomena. National data can include many measures, cover large spatial areas, and span broad environmental gradients. Because of these characteristics, analyses of these data can yield accurate estimates of relationships among different lake characteristics. However, the number of samples in a national data set that is available for estimating a relationship specific to waterbodies within a smaller region, like a single state, is limited. Conversely, state monitoring data provide intensive sampling of lakes within a smaller area, but these data span a narrower range of conditions and may only include a subset of relevant measurements. Here, a Bayesian network model is described that represents the causal linkages between observations of chlorophyll a concentration, cyanobacterial biovolume, and microcystin concentration. This network model was fit to national data and provided a context for modeling observations of chlorophyll a and microcystin collected from lakes in Iowa. Using the knowledge inherent in the national network model improved the accuracy of predictions of microcystin concentrations in Iowa compared to a model based only on Iowa data.

Application of Bayesian network including Microcystis morphospecies for microcystin risk assessment in three cyanobacterial bloom-plagued lakes, China

Microcystis spp., which occur as colonies of different sizes under natural conditions, have expanded in temperate and tropical freshwater ecosystems and caused seriously environmental and ecological problems. In the current study, a Bayesian network (BN) framework was developed to access the probability of microcystins (MCs) risk in large shallow eutrophic lakes in China, namely, Taihu Lake, Chaohu Lake, and Dianchi Lake. By means of a knowledge-supported way, physicochemical factors, Microcystis morphospecies, and MCs were integrated into different network structures. The sensitive analysis illustrated that Microcystis aeruginosa biomass was overall the best predictor of MCs risk, and its high biomass relied on the combined condition that water temperature exceeded 24 °C and total phosphorus was above 0.2 mg/L. Simulated scenarios suggested that the probability of hazardous MCs (≥1.0 μg/L) was higher under interactive effect of temperature increase and nutrients (nitrogen and phosphorus) imbalance than that of warming alone. Likewise, data-driven model development using a naïve Bayes classifier and equal frequency discretization resulted in a substantial technical performance (CCI = 0.83, K = 0.60), but the performance significantly decreased when model excluded species-specific biomasses from input variables (CCI = 0.76, K = 0.40). The BN framework provided a useful screening tool to evaluate cyanotoxin in three studied lakes in China, and it can also be used in other lakes suffering from cyanobacterial blooms dominated by Microcystis.