Study site
The Oder is an 866 km long lowland river with a catchment of 118,890 km2 in the Czech Republic, Poland, and Germany. While the upper Oder is impounded over 204 km length and fragmented by 25 weirs, its middle and lower course from river-km 300 is free flowing without any barrier to the Baltic Sea and includes several regions of extraordinary importance for ecological conservation. The Oder is highly eutrophic and has been affected by salt emissions from coal and ore mining in the upper catchment (Upper Silesia, 23). Brines from mining are partly stored in large basins with managed outlets to the Oder.
Sample collection and field measurements
Nine grab water samples were collected with rope and bucket from the free-flowing center of the Oder River, using bridges at seven sites between August 16 to 22, 2022 (Table S1). One additional water sample was obtained from a sturgeon rearing facility at Friedrichsthal (August 12th) and one was provided by the Brandenburg Office for the Environment (Frankfurt, August 15th). In the bucket samples, water temperature (T), pH value, electric conductivity (EC) and dissolved oxygen concentration and saturation were determined with a hand-held multiprobe (WTW multi 3630i). Subsamples for general water chemistry were filled in 1L HDPE bottles, and subsamples for Prymnesium identification and prymnesin analysis were filled in 1L borosilicate glass bottles. These samples were brought to the lab within 2 hours after retrieval and processed, conserved, and stored for further analysis.
Water quality analyses
Dissolved nutrients ammonium (NH4-N), nitrate (NO3-N), and soluble reactive phosphorus (SRP) were determined by flow segmented analysis (SEAL AA3) in membrane filtered (0.45µm) and acidified (2M HCl) samples. Total phosphorus (TP) was determined in original samples by molybdenum blue colorimetry after wet-chemical digestion with hydrogen peroxide and sulfuric acid. Total carbon (TC) and total nitrogen (TN) were determined in unfiltered and acidified (2M HCl) samples by catalytic oxidation and infrared spectroscopy and chemiluminescence. Sulfate (SO42-) and chloride (Cl–) were determined by ion chromatography, conductivity detection after chemical suppression (Metrohm CompactIC) in membrane filtered (0.2µm) samples. Sodium (Na), potassium (K), calcium (Ca) and magnesium (Mg) were determined in membrane filtered (0.45µm) and acidified (2M HCl) samples by inductively-coupled plasma optical emission spectroscopy.
Analyses of the phytoplankton dynamics
Phytoplankton species composition was analysed in living and in Lugol fixed subsamples. Species abundance was enumerated by inverted microscopy following 38. Species-specific cell volumes were estimated from measurements of all dimensions from at least 20 cells per species, assuming simple geometric shapes. Other subsamples were filtered onto GF/F glass-fibre filters which were immediately frozen at -80 °C. Pigment analyses were performed by HPLC after freeze-drying and extraction with DMF. For details of the chlorophyll-a analyses see 39.
DNA extraction and genotypic characterization
Prymnesium were genotyped from water samples using Sanger sequencing of the nuclear ITS region, and DNA metabarcoding with next-generation sequencing of the ITS-1 region as follows. Water samples were taken on three different days (12, 15, 19 August 2022) for a total of 7 samples from 5 sites (Küstrin, Frankfurt (Oder), Hohenwutzen, Schwedt, Friedrichsthal). DNA was extracted from filters using a DNeasyPowerSoil Pro kit (Qiagen, Hilden, Germany). We also genotyped Prymnesium from three tissue samples taken from dead juvenile sturgeon (Acipenser oxyrinchus) that had died after direct exposure to river water on 12 August 2022 in a restocking tank of a rearing facility situated next to the river near Friedrichsthal. These individuals belonged to the German conservation and species restoration program of the Oder River 40. We haphazardly selected three individual sturgeons of ca. 4 cm body length, removed ca. 0.25 g of gill tissue, extracted DNA using the DNeasy Blood and Tissue Kit (Qiagen), and Sanger-sequenced the ITS region.
For Sanger sequencing, PCR (95°C for 2 min; 38 cycles of 95°C for 2 min, 69°C for 1 min, 72°C for 1 min; 72°C for 5 min) reactions contained 5 μL 5x AllTaq buffer (Qiagen), 0.5 μL dNTPs (Agilent), 1.25 μL of each primer, 0.5 μL of AllTaq (Qiagen), and 2 μL DNA (concentration reaching from 10-88 ng/μL), and 14.5 μL RO-filtered water. We newly designed Prymnesium-specific primers (Prym_ITS_F: 5'_CCGGTCTTTCCACCCACCA_3'; Prym_ITS_R: 5'_GCCCACCGGTACGCCTCG_3') using published sequences (GenBank accessions MK091108-MK091133 25. This approach was also applied to minced gills samples of dead juvenile sturgeon (see above) to avoid potential interaction with the sturgeon genomic DNA. In this case, we also tested dilutions of 1:10, and 1: 100 as templates. Products were purified and sequenced in both directions using the same primers at Eurofins Genomics, manually edited, and aligned with the Prymnesium sequences from 25. ITS gene trees for simple sequence assignment were reconstructed using PhyML 41 with neighbor-joining and 100 bootstrap iterations. GenBank Accession numbers XXXXXX will be added upon acceptance of the paper.
Next-generation sequencing was carried out on an i7 Hybrid liquid-handling robot (Beckmann-Coulter) at the Berlin center for Genomics in Biodiversity Research. PCR reactions (98°C for 60 s; 30 cycles of 98°C for 10 s, 67°C for 20 s, 72°C for 10 s; 72°C for 2 min) included 10 ng template DNA, 12 μL Q5 High-Fidelity Master Mix (NEB), 1.25 μL each of primers that were designed for the ITS-1 region using PriSeT 42, and RO-filtered water for a total reaction volume of 25 μL. PCR products were cleaned using a magnetic bead protocol (AgencourtAMPure XP, Beckman Coulter, Indianapolis, IN, USA). DNA concentration was measured using the QuantiFluor® dsDNA System (Promega, Madison, WI, USA), and PCR products all were normalized to a concentration of 5 ng μL−1. An indexing PCR reaction (95°C for 2 min; 8 cycles of 95°C for 20 s, 52°C for 30 s, 72°C for 30 s; 72°C for 3 min) was used to add unique 12-bp inline sequence barcodes (Nextera Index Kit, Illumina, San Diego, CA, USA) to each sample, using 10 μL of target PCR product and with 5 μL reaction buffer (Herculase II Fusion DNA Polymerase, Agilent), 0.25 μL dNTPs (Agilent), 0.625 μL each of index primers P5 and P7 (Nextera Index Kit, Illumina), 0.25 μL polymerase (Herculase II Fusion DNA Polymerase, Agilent), 1 μL DMSO, and RO-filtered water for a total reaction volume of 25 μL. PCR products were purified and quantified as above. Samples were then pooled in equimolar amounts and sequenced on an Illumina MiSeq using a v3 sequencing kit (600 cycles). Negative controls were included as part of all PCR reactions and were sequenced in the same run as the regular samples. Raw sequence data (fastq.gz files) are available at the Sequence Read Archive (BioProject accession number ########### will be inserted upon paper acceptance).
Raw sequence data were analyzed using the DADA2 v 1.28.0 package 43 in R (Team R, 2023). The complete R script is provided as a supplementary file (.R). Taxonomic assignment of amplicon sequence variants (ASVs) was performed using the assignTaxonomy function with default parameters in DADA2, and the Protist Ribosomal Reference (PR2 version 4.12.0) database 44.
Prymnesin analysis
Extraction of the biomass on filter samples or the water samples was performed as described in26 with small modifications as described in 45. In brief, the filters containing the biomass were transferred to 15 mL polypropylene tubes and were extracted twice with 12 mL HPLC-grade methanol in the ultrasonic bath for 30 min. The samples were centrifuged at 3220 g for 10 min and the supernatants were combined and dried using a CentriVap Benchtop Vacuum Concentrator coupled to a CentriVapColdtrap (both Labconco Corporation, Kansas City, MO/USA) at 10 °C. For reconstitution 0.5 mL MeOH:H2O (90:10, v/v) was used. The water samples with or without biomass (90 mL each) were extracted by liquid-liquid extraction using pure 2-butanol in the following way: First, 3.42 g of NaCl were added to the water samples with or without biomass to increase the extraction efficiency of prymnesins. Thereafter, in the first extraction step 40 mL solvent and in the two consecutive extractions 20 mL each were used. The resulting three organic phases were combined, washed twice with 20 mL H2O and dried down as described above for the biomass samples. Finally, the samples were reconstituted in 0.5 mL MeOH:H2O (90:10, v/v). The presence and identity of the prymnesins were performed by liquid chromatography-mass spectrometric measurements using a Vanquish HPLC-system (Thermo Scientific, Waltham, MA/USA) coupled to a timsTOF flex system (Bruker Corporation, Billerica, MA/USA). The mass spectrometer was operated in positive electrospray ionization mode and data were collected from m/z 100 to 2000. A Kinetex F5 (2.1 × 100 mm, 2.6 µm, Phenomenex, Aschaffenburg, Germany) column equipped with a guard column of the same type was used together with a water-acetonitrile gradient both containing 0.1% formic acid and 1 mM ammonium formate. The instrument was controlled with Qtof Control Version 6.2 (Build 4.0), Compass HyStar 6.0 (Version 6.0.30.0) (both Bruker) and Chromeleon 7.3 (Thermo Scientific). The estimation of the prymnesin concentrations were performed with the indirect method described in 26 using the AccQ-Fluor reagent WAT (Waters Corporation, Milford, MA/USA) and the mycotoxins fumonisin B1 and B2 (RomerLabs, Tulln, Austria) for external calibration. A 1200 series HPLC system (Agilent Technologies, Waldbronn, Germany) was used together with an Agilent Poroshell 120 EC-C18 column (2.1 × 50 mm, 2.7 µm) and a water-acetonitrile gradient (both with 0.1% formic acid). Data evaluation was performed with Bruker Compass DataAnalysis Version 5.3. and Excel®Microsoft Corporation. All previously detected and proposed prymnesins according to 25 were investigated, but only three B-type prymnesins could be detected in the Oder samples (Supplementary Table S2). The backbone of the identified B-type prymnesin is provided in Figure 4A, the exact location of the hexose conjugates could not be determined with the applied method.
Analysis of Sentinel-2 MSI data
Chl-a concentration were derived from Sentinel-2 MSI satellite data using ESA’s Sentinel Toolbox SNAP. The Sentinel-2 L1C products were resampled to 20m spatial resolution for all spectral bands by applying the S2Rsampling processor. Water pixels were separated from land, cloud, and cloud shadow pixels by using the Idepix processor 46. Atmospheric correction and in-water retrieval were performed by applying the C2RCC processor which relies on a large database 47,48. The processor offers neural networks trained for different water types. Version C2X-COMPLEX was applied.
Extraction of river chlorophyll concentration profiles
To calculate the Chl-a profiles along the length of the river network, data derived from Sentinel-2 (Figure S3) were matched to the river network (INSPIRE, 49). The raw data were transformed into a connected graph and a spatial network to match Chl-a and geographic position. All calculations were done in R (v4.3.1),50 using the packages sf (v1.0-14) 51, sfnetworks (v0.6.3) 52, and igraph (v1.5.1) 53 for the network analysis. River discharge data was obtained from the European Flood Awareness System (EFAS), the EFAS Historical Reanalysis Data version 4.0 was used to estimate river discharge 54. Measured river discharge data was obtained from Brandenburg State Office for the Environment. The scripts of this data analysis are available in a permanent repository under https://doi.org/10.5281/zenodo.8343700. As a proxy of the total biomass of the algal bloom, the area below the Chl-a peak was estimated as the cumulated product of Chl-a concentration and river length between two adjacent data points. The area was multiplied by the actual discharge to estimate a relative algal load.