A critical review of model construction and performance for nowcast systems for faecal contamination in recreational beaches.

Faecal contamination is a widespread environmental and public health problem on recreational beaches around the world. The implementation of predictive models has been recommended by the World Health Organization as a complement to traditional monitoring to assist decision-makers and reduce health risks. Despite several advances that have been made in the modeling of faecal coliforms, tools and algorithms from machine learning are still scarcely used in the field and their implementation in nowcast systems is delayed. Here, we perform a literature review on modeling strategies to predict faecal contamination in recreational beaches in the last two decades and the implementation of models in nowcast systems to aid management. Models constructed for surface waters of continental (lakes, rivers and streams), estuarine and marine coastal ecosystems were analyzed and compared based on performance metrics for continuous (i.e. regression; R2, Root Mean Square Error: RMSE) and categorical (i.e. classification; accuracy, sensitivity, specificity) responses. We found 67 articles matching the search criteria and 40 with information allowing to evaluate and compare predictive ability. In early 2000, Multiple Linear Regressions were common, followed by a peak of Artificial Neural Networks (ANNs) from 2010 to 2015, and the rise of Machine learning techniques, such as decision trees (CART and Random Forest) since 2015. ANNs and decision trees presented better accuracy than the remaining models. Rainfall and its lags were important predictor variables followed by water temperature. Specificity was much higher than sensitivity in all modeling strategies, which is typical in data sets where one category (e.g. closed beach) is far less common than the normal state (i.e. unbalanced data sets). We registered the implementation of statistical models in early warning systems in 6 countries, mainly by public beach quality management institutions, followed by NGOs in conjunction with universities. We identified critical steps towards improving model construction, evaluation and usage: i) the need to balance the data set previous to model training, ii) the need to separate data set in training, validation and test to perform an honest evaluation of model performance and iii) the transduction of model outputs to plain language to relevant stakeholders. Integrating into a single framework in situ monitoring, model construction and nowcasting systems could help to improve decision making systems to protect users from bathing in contaminated waters. Still the reduction of arrival of faecal coliforms to aquatic ecosystems (e.g. by improving sewage treatment systems) will be the ultimate factor in reducing health risk.

The Microcystis-microbiome interactions: origins of the colonial lifestyle.

Species of the Microcystis genus are the most common bloom-forming toxic cyanobacteria worldwide. They belong to a clade of unicellular cyanobacteria whose ability to reach high biomasses during blooms is linked to the formation of colonies. Colonial lifestyle provides several advantages under stressing conditions of light intensity, ultraviolet light, toxic substances and grazing. The progression from a single-celled organism to multicellularity in Microcystis has usually been interpreted as individual phenotypic responses of the cyanobacterial cells to the environment. Here, we synthesize current knowledge about Microcystis colonial lifestyle and its role in the organism ecology. We then briefly review the available information on Microcystis microbiome and propose that changes leading from single cells to colonies are the consequence of specific and tightly regulated signals between the cyanobacterium and its microbiome through a biofilm-like mechanism. The resulting colony is a multi-specific community of interdependent microorganisms.

Effect of hydrological modification on the potential toxicity of Microcystis aeruginosa complex in Salto Grande reservoir, Uruguay.

It is widely known that the environmental conditions caused by the construction of reservoirs favor the proliferation of toxic cyanobacteria and the formation of blooms due to the high residence time of the water, low turbidity, temperature regimes, among others. Microcystin-producing cyanobacteria such as those from the Microcystis aeruginosa complex (MAC) are the most frequently found organisms in reservoirs worldwide, being the role of the environment on microcystin production poorly understood. Here, we addressed the community dynamics and potential toxicity of MAC cyanobacteria in a subtropical reservoir (Salto Grande) located in the low Uruguay river. Samples were taken from five different sites (upstream, inside the reservoir and downstream) during contrasting seasons (summer and winter) to analyze: (i) the MAC community structure by amplicon sequencing of the phycocyanin gene spacer, (ii) the genotype diversity of microcystin-producing MAC by high resolution melting analysis of the mcyJ gene, and (iii) the abundance and mcy transcription activity of the microcystin-producing (toxic) fraction. We found that MAC diversity decreased from summer to winter but, despite the observed changes in MAC community structure, the abundance of toxic organisms and the transcription of mcy genes were always higher inside the reservoir, regardless of the season. Two different genotypes of toxic MAC were detected inside the reservoir, one associated with low water temperature (15 °C) and one thriving at high water temperature (31 °C). These findings indicate that the environmental conditions inside the reservoir reduce community diversity while promoting the proliferation of toxic genotypes that actively transcribe mcy genes, whose relative abundance will depend on the water temperature.

Rise of toxic cyanobacterial blooms is promoted by agricultural intensification in the basin of a large subtropical river of South America.

Toxic cyanobacterial blooms are globally increasing with negative effects on aquatic ecosystems, water use and human health. Blooms' main driving forces are eutrophication, dam construction, urban waste, replacement of natural vegetation with croplands and climate change and variability. The relative effects of each driver have not still been properly addressed, particularly in large river basins. Here, we performed a historical analysis of cyanobacterial abundance in a large and important ecosystem of South America (Uruguay river, ca 1900 km long, 365,000 km2 basin). We evaluated the interannual relationships between cyanobacterial abundance and land use change, river flow, urban sewage, temperature and precipitation from 1963 to the present. Our results indicated an exponential increase in cyanobacterial abundance during the last two decades, congruent with an increase in phosphorus concentration. A sharp shift in the cyanobacterial abundance rate of increase after the year 2000 was identified, resulting in abundance levels above public health alert since 2010. Path analyses showed a strong positive correlation between cyanobacteria and cropland area at the entire catchment level, while precipitation, temperature and water flow effects were negligible. Present results help to identify high nutrient input agricultural practices and nutrient enrichment as the main factors driving toxic bloom formation. These practices are already exerting severe effects on both aquatic ecosystems and human health and projections suggest these trends will be intensified in the future. To avoid further water degradation and health risk for future generations, a large-scale (transboundary) change in agricultural management towards agroecological practices will be required.

Las floraciones de cianobacterias tóxicas vienen aumentando drásticamente a nivel mundial con efectos negativos en los ecosistemas acuáticos, los usos del agua y la salud humana. Los principales mecanismos promotores de las floraciones son la eutrofización, la construcción de represas, la contaminación con residuos urbanos, la pérdida de vegetación natural y el cambio y la variabilidad climáticos. Los efectos relativos de cada determinante aún no se han abordado adecuadamente, particularmente en las grandes cuencas fluviales de América del Sur. En este trabajo, realizamos un análisis histórico de la abundancia de cianobacterias en un gran e importante ecosistema de América del Sur (el Río Uruguay, c.a. 1.900 km de largo, cuenca de 365.000 km2). Evaluamos las relaciones entre la abundancia de cianobacterias y el cambio en los usos del suelo, el caudal de los ríos, la contaminación urbana, la temperatura y la precipitación desde 1963 hasta el presente. Nuestros resultados evidencian un aumento exponencial en la abundancia de cianobacterias durante las últimas dos décadas, de forma congruente con el aumento en la concentración de fósforo en agua. Fue identificado además, un cambio brusco en la tasa de aumento de la abundancia de cianobacterias después del año 2000, lo que resultó en niveles de alerta por encima de riesgo para la salud pública desde 2010. Los análisis estadísticos indicaron una fuerte y positiva correlación entre las cianobacterias y el área de cultivo en la cuenca, mientras que la precipitación, la temperatura y el flujo de agua fueron insignificantes. Estos resultados contribuyen a identificar que las prácticas agrícolas con alto aporte de nutrientes y el enriquecimiento de nutrientes son los principales impulsores de la formación de floraciones tóxicas. Estas prácticas ya están teniendo graves efectos en los ecosistemas acuáticos y la salud humana y las proyecciones sugieren que se intensificarán en el futuro. Para evitar una mayor degradación de la calidad del agua y el incremento de los riesgos para la salud de las generaciones futuras, se requerirá un cambio a gran escala (transfronterizo) en la gestión agrícola hacia prácticas agroecológicas.

Distance decay 2.0 - A global synthesis of taxonomic and functional turnover in ecological communities.

Aim: Understanding the variation in community composition and species abundances (i.e., ß-diversity) is at the heart of community ecology. A common approach to examine ß-diversity is to evaluate directional variation in community composition by measuring the decay in the similarity among pairs of communities along spatial or environmental distance. We provide the first global synthesis of taxonomic and functional distance decay along spatial and environmental distance by analysing 148 datasets comprising different types of organisms and environments. Location: Global. Time period: 1990 to present. Major taxa studied: From diatoms to mammals. Method: We measured the strength of the decay using ranked Mantel tests (Mantel r) and the rate of distance decay as the slope of an exponential fit using generalized linear models. We used null models to test whether functional similarity decays faster or slower than expected given the taxonomic decay along the spatial and environmental distance. We also unveiled the factors driving the rate of decay across the datasets, including latitude, spatial extent, realm and organismal features. Results: Taxonomic distance decay was stronger than functional distance decay along both spatial and environmental distance. Functional distance decay was random given the taxonomic distance decay. The rate of taxonomic and functional spatial distance decay was fastest in the datasets from mid-latitudes. Overall, datasets covering larger spatial extents showed a lower rate of decay along spatial distance but a higher rate of decay along environmental distance. Marine ecosystems had the slowest rate of decay along environmental distances. Main conclusions: In general, taxonomic distance decay is a useful tool for biogeographical research because it reflects dispersal-related factors in addition to species responses to climatic and environmental variables. Moreover, functional distance decay might be a cost-effective option for investigating community changes in heterogeneous environments.

Genotyping and Multivariate Regression Trees Reveal Ecological Diversification within the Microcystis aeruginosa Complex along a Wide Environmental Gradient.

Addressing the ecological and evolutionary processes underlying biodiversity patterns is essential to identify the mechanisms shaping community structure and function. In bacteria, the formation of new ecologically distinct populations (ecotypes) is proposed as one of the main drivers of diversification. New ecotypes arise when mutations in key functional genes or acquisition of new metabolic pathways by horizontal gene transfer allow the population to exploit new resources, permitting their coexistence with the parental population. We previously reported the presence of microcystin-producing organisms of the Microcystis aeruginosa complex (toxic MAC) through an 800-km environmental gradient ranging from freshwater to estuarine-marine waters in South America. We hypothesize that the success of toxic MAC in such a gradient is due to the existence of very closely related populations that are ecologically distinct (ecotypes), each specialized to a specific arrangement of environmental variables. Here, we analyzed toxic MAC genetic diversity through quantitative PCR (qPCR) and high-resolution melting analysis (HRMA) of a functional gene (mcyJ, microcystin synthetase cluster). We explored the variability of the mcyJ gene along the environmental gradient by multivariate classification and regression trees (mCART). Six groups of mcyJ genotypes were distinguished and associated with different combinations of water temperature, conductivity, and turbidity. We propose that each mcyJ variant associated with a defined environmental condition is an ecotype (or species) whose relative abundances vary according to their fitness in the local environment. This mechanism would explain the success of toxic MAC in such a wide array of environmental conditions. IMPORTANCE Organisms of the Microcystis aeruginosa complex form harmful algal blooms (HABs) in nutrient-rich water bodies worldwide. MAC HABs are difficult to manage owing to the production of potent toxins (microcystins) that resist water treatment. In addition, the role of microcystins in the ecology of MAC organisms is still elusive, meaning that the environmental conditions driving the toxicity of the bloom are not clear. Furthermore, the lack of coherence between morphology-based and genomic-based species classification makes it difficult to draw sound conclusions about when and where each member species of the MAC will dominate the bloom. Here, we propose that the diversification process and success of toxic MAC in a wide range of water bodies involves the generation of ecotypes, each specialized in a particular niche, whose relative abundance varies according to its fitness in the local environment. This knowledge can improve the generation of accurate prediction models of MAC growth and toxicity, helping to prevent human and animal intoxication.

Machine learning methods for imbalanced data set for prediction of faecal contamination in beach waters.

Predicting water contamination by statistical models is a useful tool to manage health risk in recreational beaches. Extreme contamination events, i.e. those exceeding normative are generally rare with respect to bathing conditions and thus the data is said to be imbalanced. Modeling and predicting those rare events present unique challenges. Here we introduce and evaluate several machine learning techniques and metrics to model imbalanced data and evaluate model performance. We do so by using a) simulated data-sets and b) a real data base with records of faecal coliform abundance monitored for 10 years in 21 recreational beaches in Uruguay (N ≈ 19000) using in situ and meteorological variables. We discuss advantages and disadvantages of the methods and provide a simple guide to perform models for a general audience. We also provide R codes to reproduce model fitting and testing. We found that most Machine Learning techniques are sensitive to imbalance and require specific data pre-treatment (e.g. upsampling) to improve performance. Accuracy (i.e. correctly classified cases over total cases) is not adequate to evaluate model performance on imbalanced data set. Instead, true positive rates (TPR) and false positive rates (FPR) are recommended. Among the 52 possible candidate algorithms tested, the stratified Random forest presented the better performance improving TPR in 50% with respect to baseline (0.4) and outperformed baseline in the evaluated metrics. Support vector machines combined with upsampling method or synthetic minority oversampling technique (SMOTE) performed well, similar to Adaboost with SMOTE. These results suggests that combining modeling strategies is necessary to improve our capacity to anticipate water contamination and avoid health risk.

Rapid freshwater discharge on the coastal ocean as a mean of long distance spreading of an unprecedented toxic cyanobacteria bloom.

Cyanobacterial toxic blooms are a worldwide problem. The Río de la Plata (RdlP) basin makes up about one fourth of South America areal surface, second only to the Amazonian. Intensive agro-industrial land use and the construction of dams have led to generalized eutrophication of main tributaries and increased the intensity and duration of cyanobacteria blooms. Here we analyse the evolution of an exceptional bloom at the low RdlP basin and Atlantic coast during the summer of 2019. A large array of biological, genetic, meteorological, oceanographic and satellite data is combined to discuss the driving mechanisms. The bloom covered the whole stripe of the RdlP estuary and the Uruguayan Atlantic coasts (around 500 km) for approximately 4 months. It was caused by the Microcystis aeruginosa complex (MAC), which produces hepatotoxins (microcystin). Extreme precipitation in the upstream regions of Uruguay and Negro rivers' basins caused high water flows and discharges. The evolution of meteorological and oceanographic conditions as well as the similarity of organisms' traits in the affected area suggest that the bloom originated in eutrophic reservoirs at the lower RdlP basin, Salto Grande in the Uruguay river, and Negro river reservoirs. High temperatures and weak Eastern winds prompted the rapid dispersion of the bloom over the freshwater plume along the RdlP northern and Atlantic coasts. The long-distance rapid drift allowed active MAC organisms to inoculate freshwater bodies from the Atlantic basin, impacting environments relevant for biodiversity conservation. Climate projections for the RdlP basin suggest an increase in precipitation and river water flux, which, in conjunction with agriculture intensification and dams' construction, might turn this extraordinary event into an ordinary situation.

Morphology captures toxicity in Microcystis aeruginosa complex: Evidence from a wide environmental gradient✰.

Blooms of the Microcystis aeruginosa complex (MAC) consist of mixtures of toxin-producing and non-toxin-producing populations, but the environmental conditions that determine their relative abundance and shift are not clear. Morphological traits reflect the responses of MAC organisms to environmental changes, thus they could be useful to improve the predictability of the abundance of both toxic and nontoxic populations. In this work, the response of MAC toxic populations to environmental conditions and their relationship with morphology (size of organisms) were investigated in different water bodies (reservoir, river, and estuary) covering wide salinity (0-33) and temperature (10-36 °C) gradients. Sub-surface water samples were collected and divided into 4 size classes (mesh size 〈20 µm, 20-60 µm, 60-150 µm and〉 150 µm) and three toxicity proxies were assessed (mcyE gene and transcripts copy numbers and microcystin concentration) for each size-class. For all the size-classes, the logarithm of the number of mcyE gene copies per sample was proportional to the logarithm of the corresponding biovolume fraction, showing that MAC biovolume is a good indicator of toxicity potential. When toxicity was analyzed through mcyE transcript abundance and microcystin concentration, the largest size fraction (>150 µm) showed the highest toxicity values of both proxies. Nevertheless, mcyE transcription and toxin production per cell were higher in the colonies retained in the 60 to 150 µm size fractions, followed by single cells (<20 µm). At the reservoir, where environmental variability is low, the total abundance of mcyE gene copies was significantly explained by MAC biovolume, regardless of the environmental conditions. However, when data from the reservoir to the estuary were modeled, biovolume and temperature (with a minor contribution of salinity and wind intensity) were selected in the best models. According to these results, the size distribution of MAC biovolume appears as a good predictor of active toxin production, being the colonies in the 60-150 µm size fraction good indicators of higher toxicity. These results can be used to predict MAC toxicity based on the size structure of the community.

Morphology-based differences in the thermal response of freshwater phytoplankton.

The thermal response of maximum growth rate in morphology-based functional groups (MBFG) of freshwater phytoplankton is analysed. Contrasting an exponential Boltzmann-Arrhenius with a unimodal model, three main features were evaluated: (i) the activation energy of the rise (Er), (ii) the presence of a break in the thermal response and (iii) the activation energy of the fall (Ef). The whole dataset (N = 563) showed an exponential increase (Er ∼ 0.5), a break around 24°C and no temperature dependence after the breakpoint (Ef = 0). Contrasting thermal responses among MBFG were found. All groups showed positive activation energy (Er > 0), four showed no evidence of decline in growth rate (temperature range = 0-35°C) and two presented a breakpoint followed by a sharp decrease in growth rate. Our results evidenced systematic differences between MBFG in the thermal response and a coherent response significantly related to morphological traits other than size (i.e. within MBFG). These results provide relevant information for water quality modelling and climate change predictions.

A multilevel trait-based approach to the ecological performance of Microcystis aeruginosa complex from headwaters to the ocean.

The Microcystis aeruginosa complex (MAC) clusters cosmopolitan and conspicuous harmful bloom-forming cyanobacteria able to produce cyanotoxins. It is hypothesized that low temperatures and brackish salinities are the main barriers to MAC proliferation. Here, patterns at multiple levels of organization irrespective of taxonomic identity (i.e. a trait-based approach) were analyzed. MAC responses from the intracellular (e.g. respiratory activity) to the ecosystem level (e.g. blooms) were evaluated in wide environmental gradients. Experimental results on buoyancy and respiratory activity in response to increased salinity (0-35) and a literature review of maximum growth rates under different temperatures and salinities were combined with field sampling from headwaters (800km upstream) to the marine end of the Rio de la Plata estuary (Uruguay-South America). Salinity and temperature were the major variables affecting MAC responses. Experimentally, freshwater MAC cells remained active for 24h in brackish waters (salinity=15) while colonies increased their flotation velocity. At the population level, maximum growth rate decreased with salinity and presented a unimodal exponential response with temperature, showing an optimum at 27.5°C and a rapid decrease thereafter. At the community and ecosystem levels, MAC occurred from fresh to marine waters (salinity 30) with a sustained relative increase of large mucilaginous colonies biovolume with respect to individual cells. Similarly, total biomass and, specific and morphological richness decreased with salinity while blooms were only detected in freshwater both at high (33°C) and low (11°C) temperatures. In brackish waters, large mucilaginous colonies presented advantages under osmotic restrictive conditions. These traits values have also been associated with higher toxicity potential. This suggest salinity or low temperatures would not represent effective barriers for the survival and transport of potentially toxic MAC under likely near future scenarios of increasing human impacts (i.e. eutrophication, dam construction and climate change).

Dynamics of toxic genotypes of Microcystis aeruginosa complex (MAC) through a wide freshwater to marine environmental gradient.

Bloom-forming species belonging to Microcystis aeruginosa complex (MAC) are the most commonly reported worldwide. MAC blooms are composed by toxic and non-toxic genotypes and the environmental conditions favouring the dominance of toxic genotypes are still a matter of debate among the scientific community. In this study, we evaluated the distribution of toxic MAC genotypes along a seasonal cycle and over an environmental gradient spanning 800km, from a eutrophic freshwater reservoir in Río Uruguay to marine water in the outer limit of Río de la Plata. Abundance of four mcy genes, mcyB, mcyD, mcyE and mcyJ was determined by qPCR and used as a proxy of abundance of toxic MAC genotypes. All the mcy genes were detected through the seasonal cycle at all sampling sites, being systematically higher in the freshwater reservoir and decreasing towards the marine site. The highest toxic genotype abundance was found during the austral summer months. According to generalized linear regressions and random forest models, temperature and conductivity were the most relevant explanatory variables. This suggests that although toxic MAC genotypes grow optimally in freshwater, they are also able to tolerate the high-salinity and low temperature conditions found in estuarine and marine waters. This ability to resist harsh conditions impose a health risk and a management challenge. To our knowledge, this is the first report addressing several mcy genes in a broad gradient that includes a wide array of different environmental conditions.

A unimodal species response model relating traits to environment with application to phytoplankton communities.

In this paper we attempt to explain observed niche differences among species (i.e. differences in their distribution along environmental gradients) by differences in trait values (e.g. volume) in phytoplankton communities. For this, we propose the trait-modulated Gaussian logistic model in which the niche parameters (optimum, tolerance and maximum) are made linearly dependent on species traits. The model is fitted to data in the Bayesian framework using OpenBUGS (Bayesian inference Using Gibbs Sampling) to identify according to which environmental variables there is niche differentiation among species and traits. We illustrate the method with phytoplankton community data of 203 lakes located within four climate zones and associated measurements on 11 environmental variables and six morphological species traits of 60 species. Temperature and chlorophyll-a (with opposite signs) described well the niche structure of all species. Results showed that about 25% of the variance in the niche centres with respect to chlorophyll-a were accounted for by traits, whereas niche width and maximum could not be predicted by traits. Volume, mucilage, flagella and siliceous exoskeleton are found to be the most important traits to explain the niche centres. Species were clustered in two groups with different niches structures, group 1 high temperature-low chlorophyll-a species and group 2 low temperature-high chlorophyll-a species. Compared to group 2, species in group 1 had larger volume but lower surface area, had more often flagella but neither mucilage nor siliceous exoskeleton. These results might help in understanding the effect of environmental changes on phytoplankton community. The proposed method, therefore, can also apply to other aquatic or terrestrial communities for which individual traits and environmental conditioning factors are available.

What drives the distribution of the bloom-forming cyanobacteria Planktothrix agardhii and Cylindrospermopsis raciborskii?

The cyanobacteria Planktothrix agardhii and Cylindrospermopsis raciborskii are bloom-forming species common in eutrophic freshwaters. These filamentous species share certain physiological traits which imply that they might flourish under similar environmental conditions. We compared the distribution of the two species in a large database (940 samples) covering different climatic regions and the Northern and Southern hemispheres, and carried out laboratory experiments to compare their morphological and physiological responses. The environmental ranges of the two species overlapped with respect to temperature, light and total phosphorus (TP); however, they responded differently to environmental gradients; C. raciborskii biovolume changed gradually while P. agardhii shifted sharply from being highly dominated to a rare component of the phytoplankton. As expected, P. agardhii dominates the phytoplankton with high TP and low light availability conditions. Contrary to predictions, C. raciborskii succeeded in all climates and at temperatures as low as 11 °C. Cylindrospermopsis raciborskii had higher phenotypic plasticity than P. agardhii in terms of pigments, individual size and growth rates. We conclude that the phenotypic plasticity of C. raciborskii could explain its ongoing expansion to temperate latitudes and suggest its future predominance under predicted climate-change scenarios.

Emergent neutrality drives phytoplankton species coexistence.

The mechanisms that drive species coexistence and community dynamics have long puzzled ecologists. Here, we explain species coexistence, size structure and diversity patterns in a phytoplankton community using a combination of four fundamental factors: organism traits, size-based constraints, hydrology and species competition. Using a 'microscopic' Lotka-Volterra competition (MLVC) model (i.e. with explicit recipes to compute its parameters), we provide a mechanistic explanation of species coexistence along a niche axis (i.e. organismic volume). We based our model on empirically measured quantities, minimal ecological assumptions and stochastic processes. In nature, we found aggregated patterns of species biovolume (i.e. clumps) along the volume axis and a peak in species richness. Both patterns were reproduced by the MLVC model. Observed clumps corresponded to niche zones (volumes) where species fitness was highest, or where fitness was equal among competing species. The latter implies the action of equalizing processes, which would suggest emergent neutrality as a plausible mechanism to explain community patterns.