Cyanobacterial blooms in subtropical riverine and estuarine ecosystems of South America

Water quality impairment caused by toxic cyanobacterial blooms is a growing global concern adversely affecting the biodiversity and functioning of aquatic ecosystems, which can disrupt recreation and human health. Recent studies indicate that factors such as eutrophication, dam construction, and climate change are likely to increase the frequency and intensity of these blooms in aquatic ecosystems worldwide. This trend raises concerns in the subtropical South America (SA) region, where the pampas ecosystem has registered a sustained increase in the surface used by agroindustrial activities which leads to eutrophication of the Uruguay River (UR) and the Río de la Plata estuary (RdlP) ecosystems. The UR-RdlP system is crucial for recreational activities and serves as an essential water source. Historical monitoring data indicate that currently, toxic blooms are often documented in the UR and transported downstream to the RdlP (Kruk et al., 2017; Martínez de la Escalera et al., 2017). 

In this context, it is imperative to develop comprehensive and coherent reviewed datasets to analyze the spatio-temporal dynamics of toxic cyanobacterial blooms effectively. Despite the availability of public information, its accessibility and suitability for analysis are not always guaranteed. Therefore, establishing and maintaining comprehensive long-term databases in ecosystems frequented for recreational purposes is crucial for studying the mechanisms associated with bloom formation and predicting human health risks. Here, we provide historical records (1963-2022) and indices of toxic cyanobacterial blooms at ca. 80 sites in the subtropical region along the Uruguay River (UR) and Río de la Plata (RdlP). The data compilation process involved gathering dispersed information from open sources,  research projects, reports from multiple water quality monitoring programs, and collaborative efforts with research institutions in the country and the region. Data was checked for consistency and included geospatial data on cyanobacterial cell abundance, microcystin concentration, chlorophyll-a concentration, and risk levels from field samples combined with relevant environmental, land use, and climatic variables. This included in-situ measured environmental variables (e.g., water temperature, salinity, turbidity, conductivity) and regional climate and hydrology information (e.g., precipitation and flow rates), as well as land use patterns in the UR basin (e.g., crops, forestation, grasslands).

A fundamental contribution of this dataset lies in the consolidation and integration of variables reviewed from different sources, facilitating its utilization to evaluate the frequency and intensity of cyanobacterial blooms in a framework of productive intensification and climate change, to analyze the causes and effects of cyanobacterial blooms in riverine and estuarine recreational beaches and their relation with human health risks, to understand the historical dynamic of water quality experienced by users of these aquatic ecosystems, and to model and improve early warning and national monitoring systems, helping to mitigate potential public health risks. In short, various studies utilizing the provided dataset reveal the following trends: over the temporal analysis, there is a sustained increase in cyanobacteria abundance from 1960 to the present, particularly marked by an exponential growth around the year 2000 (Kruk et al., 2023). This shift is associated with changes in land use, notably the transition to industrial crops (Kruk et al., 2023). Cyanobacteria organisms and their bloom frequency of occurrence increased also in estuarine waters (Martinez de la Escalera et al., 2017; Kruk et al., 2017). Elevated salinity selects larger cyanobacterial organisms with high toxicity (Kruk et al., 2019). Cyanotoxin levels in UR and RdlP are significantly high, posing substantial public health risks, especially to vulnerable populations (Kruk et al., 2019).

We highlight the potential of this dataset to explore the interplay between environmental factors, anthropogenic changes, and cyanobacterial dynamics at recreational beaches over an extended historical period in which many relevant transitions were recorded that promoted the rise and intensification of harmful algal blooms. Its significance extends to aiding researchers and healthcare professionals in establishing specific conditions for beach water quality management.

Methods
Study area
The Uruguay River (UR) is one of the largest rivers in South America (SA). Its basin spans from 28°S to 37°S, covering a vast area of ca. 365,000 km2 and a linear extension of 1,838 km, of which ~540 km is the border between Argentina and Uruguay. At 31°S, the Salto Grande (SG) dam was built in 1974 to produce electricity (length ~100 km, average depth 6.4 m) and frequently presents toxic cyanobacterial blooms. At 35°S on the Atlantic coast of SA, lies the Rio de la Plata estuary (RdlP), with an extension of 325 km and a mean depth of 10 meters, draining the second largest basin of SA (3,170,000 km2) formed by the Paraná River and the UR. Within this basin, agricultural and industrial activities thrive, with approximately 15 million people living along its margins.
Sampling and data collection
Water quality, cyanobacterial blooms, and risk indices

Cyanobacteria information was compiled from publicly accessible reports and through formal requests from the Uruguayan Government Drinking Water Institution (Obras Sanitarias del Estado, OSE), the binational commission of the Rio Uruguay (Comisión Administradora del Río Uruguay, CARU), the binational Technical Commission of the Salto Grande Dam (CTM), the Municipality of Montevideo (IM), and different joint research projects from the University of the Republic in Uruguay (Centro Universitario Regional Este, Facultad de Ciencias, Instituto de Investigaciones Clemente Estable, and Laboratorio Técnológico del Uruguay).

Two datasets are presented. One corresponds to historical data on cyanobacterial abundance (cells mL⁻¹) from 1963 to 2019, primarily originated by OSE during water quality control campaigns at five sites adjacent to water treatment plants along the Uruguay River. The other dataset corresponds to the presence of toxic cyanobacterial blooms from 2008 to 2022, obtained through requests to environmental agencies responsible for public health and compiled from reports on the status of recreational water available on public servers. This dataset mainly consists of risk exposure indices. Both datasets were complemented with information retrieved from research projects.
During monitoring campaigns, samples were collected from subsurface water at 54 sites in the UR and 25 sites in the RdlP, to analyze the presence of cyanobacterial blooms based on cell count, microcystin concentration, and chlorophyll-a concentration. Cyanobacteria abundance was counted in inverted optical microscopy, cyanotoxins (i.e., microcystin-LR) concentrations were estimated with the Microcystins-ADDA ELISA method, and chlorophyll-a spectrophotometrically. Risk levels were assessed considering threshold values for variables indicating the presence of cyanobacteria used by environmental agencies (including cell counts, microcystin-LR concentrations, and visual inspections). Information was retrieved from the main beach water quality monitoring programs in Uruguay. In the UR, CTM monitors 16 sites along the 100 km coastline of the UR in the upper stream of the Salto Grande reservoir, while CARU monitors more than 40 sites along the entire river, extending approximately 300 km downstream from the dam, employing a comprehensive beach condition classification system. This classification is based on multiple indicators, including the presence of foam, chlorophyll-a levels, cyanobacterial cell counts, and microcystins. Exposure risk is subsequently categorized into three levels: “green” indicating low risk, “yellow” indicating moderate risk, and “red” indicating high risk and following threshold values for each variable (threshold values are provided in the readme). In the RdlP, the IM employs a visual surveillance system for cyanobacteria at 21 recreational beaches in Montevideo, categorizing cyanobacterial presence into three categories: “no” detectable colonies, low concentration with scattered “colonies”, and very high concentration with visible “foam”. To merge data retrieved from both monitoring systems (i.e., IM and CARU), and from research projects, we derived a unified three-level risk alert variable using in-situ observations from each dataset. This variable considered multiple indicators of cyanobacteria presence (i.e., cell count, chlorophyll-a and microcystin concentrations, the absence or presence of colonies, and foam). The maximum alert level was determined by identifying the highest risk level among these variables, representing the most severe scenario for public health.

Cyanobacteria data was further linked to relevant environmental variables, such as in-situ water characteristics (i.e, temperature, salinity, turbidity, nutrient concentration, etc), and time series of meteorological, climatological, and hydrological regional variables

Environmental variables

Water temperature, conductivity, dissolved oxygen, turbidity, and salinity of the water were recorded on the surface using a multiprobe. Also, data on water temperatures were measured by thermometer-equipped buoys across the UR between 2008 and 2021. The measurement of major inorganic nutrients, including Nitrogen (in the forms of nitrite, nitrate, and ammonium), and Phosphorus (as orthophosphate), followed the procedures outlined by the American Public Health Association. When there was no in-situ register of nutrient concentration, this information was supplemented with data from reports generated by the PROCON project (Water Quality and Pollution Control Program) in the UR.
For the RdlP, relevant local meteorological data were retrieved from public servers, the Uruguayan Meteorological Institute (INUMET, station Paysandú), the National Agricultural Research Institute of Uruguay (INIA, station Las Brujas), and Meteomanz.com. The recorded variables included precipitation, air temperature, and wind speed and direction; variables derived from the incorporated ones were also calculated (e.g., lagged variables and accumulated precipitation).

Hydrology, climate, and land use

Historical UR flow data (monthly and daily, average, minimum, and maximum) were acquired from the Argentine National Information System (SNIH) from 1908 to 2022 and station #3802 named Paso de los Libres upstream of Salto Grande Dam. Annual average values of maximum, minimum, and average daily temperatures and the average annual precipitation in the UR watershed from 1901 to 2022 were obtained from the historical climate database produced by the Climatic Research Unit (CRU) of the University of East Anglia and provided by the Work Bank climate knowledge portal for the watershed #252. Land use/cover maps from 2000 to 2019 for the entire UR basin were obtained from the MapBiomas Pampa project based on the use of satellite images (i.e., Landsat), and were grouped into 6 classes: natural forest, forest plantation, grasslands, and wetlands, crops and pasture, non-vegetated areas and water bodies. Data from 1985 to 2019 for the Brazilian segment of the UR basin, was used to obtain other land use categories, including pastures, soybeans, other summer crops, and a mixture of pastures and annual crops.

The integration of in-situ and regional data into a single matrix was implemented based on geographic proximity and taking into account the date in the in-situ variables to which the value corresponding to the previous day of meteorological and hydrological variables was associated. For annual data, each meteorological land use change value was repeated in all the years' observations. In this way, biological (i.e., cyanobacterial data) derived from water quality monitoring programs and research projects alongside the spatial positions of sampling sites were compiled and integrated with the environmental matrices (i.e., water characteristics, hydrological, climatological, and land use variables).