Harmful algal blooms (HABs), or the excessive growth of certain taxa of protists, cyanobacteria, or algae in water, are of growing international concern (Chorus and Welker, 2021). In freshwater, cyanobacteria harmful algal blooms (cyanoHABs) can propagate under favorable conditions and produce secondary metabolites, such as cyanotoxins, which can have adverse effects on water quality and the health of aquatic ecosystems (Brooks and others, 2016). Cyanotoxins can cause acute toxicity in mammals. Exposure to cyanotoxins through ingestion, inhalation of aerosolized water droplets, or dermal contact has the potential to produce a wide range of symptoms in humans, including liver failure, gastrointestinal distress, or, in rare cases, death (Paerl and Otten, 2013; Chorus and Welker, 2021; Patiño and others, 2023). Of the most common classes of cyanotoxins (including anatoxins, microcystins, cylindrospermopsin and saxitoxins), microcystins are the most frequently detected and researched (Chorus and Welker, 2021). Increases in the frequency, duration, and magnitude of cyanoHABs have been observed globally in recent decades. Future climate projections, including the potential warming of surface waters and changes in precipitation, indicate increasingly favorable conditions for the growth of cyanobacteria (Paerl and Huisman, 2008; O’Neil and others, 2012; Brooks and others, 2016).

Many lakes and reservoirs that are used for drinking water are located within watersheds that have documented occurrences of cyanoHABs (Miller and others, 2017; Reinl and others, 2021), including the northeastern United States (Gifford and others, 2024; Hsu and others, 2024). Although past studies of cyanoHAB development in lakes are extensive, less is known regarding cyanoHAB formation and persistence in flowing-water systems. As opposed to lakes, in rivers, there is not always a direct correlation between nutrient availability and cyanobacteria amounts (Jarvie and others, 2025). The transport of cyanobacteria and associated cyanotoxins from lakes and reservoirs during cyanoHABs has been previously documented and modeled (Graham and others, 2012; Otten and others, 2015; Schmadel and others, 2024), but the mechanisms that control cyanobacteria occurrence and growth in rivers are not well understood (Bukaveckas and others, 2018; Graham and others, 2020; Zuellig and others, 2021).

CyanoHAB monitoring often relies on analysis of discrete samples, which provide a snapshot of the phytoplankton assemblage and cyanotoxin concentrations at a specific location and time (Chorus and Welker, 2021). Aquatic ecosystems are rarely uniform in composition, and the concentration of cyanotoxins and cyanobacteria cell densities is likely to be affected by many environmental factors, including water temperature, residence time, wind speed and direction, wave action, nutrient supply, and species competition (O’Neil and others, 2012; Paerl and Otten, 2013). Because of rapidly changing environmental conditions and the high spatial and temporal variability of cyanoHABs, discrete sampling methods may be insufficient to accurately capture the dynamic nature of the blooms.

Recent advances in genetic methodologies have increased the ability to detect cyanobacteria at low concentrations and the potential for cyanotoxin production through the measurement of specific genes, such as cyanobacteria 16S ribosomal ribonucleic acid (rRNA) and cyanotoxin synthetase genes (Al-Tebrineh and others, 2012; Stelzer and others, 2013; Graham and others, 2020; Hsu and others, 2024). Deoxyribonucleic acid (DNA)-based quantitative polymerase chain reaction (qPCR) assays quantify the presence of specific genes, which indicates the potential for cyanotoxin production; however, it is important to note that the presence of the cyanotoxin synthetase genes does not guarantee cyanotoxin occurrence. The cyanobacteria must first transcribe the DNA-encoded cyanotoxin gene into messenger RNA (mRNA) and then translate the mRNA into proteins to initiate the building of the cyanotoxin (Stelzer and others, 2013). Large datasets that enable comparison among cyanobacteria assemblage, cyanotoxin concentration, and cyanobacteria genetic composition are rare, especially in riverine systems (Graham and others, 2020), and are valuable for improving understanding of the relation between cyanobacteria abundance, cyanotoxin concentrations, and the results of genetic analyses.

Solid phase adsorption toxin tracking (SPATT) is an additional alternative method of measuring cyanotoxins through passive sampler deployment that can integrate cyanotoxin concentrations over time. The SPATT samplers are used to measure the adsorption of dissolved cyanotoxins on a known amount of resin beads, with increased cyanotoxin concentrations correlated to the total time of resin exposure within the water environment (Kudela, 2011). When the samplers are retrieved after deployment, cyanotoxins are extracted from the resin, and the extract is analyzed with standard analytical methods, such as liquid chromatography-tandem mass spectrometry (LC-MS/MS; Johnston and others, 2025). The SPATT samplers are relatively low-cost, simple to build, and easy to transport, deploy, and retrieve (Negrey and others, 2019; Jaegge and others, 2026). They can be deployed in areas with limited access where collecting discrete samples is challenging or in conditions not conducive to discrete sampling, such as high-flow conditions. Passive samplers have been shown to be a reliable, sensitive, and effective time-integrated sampling approach for marine, brackish, and freshwater contexts (Kudela, 2011; Jaegge and others, 2026). Combining passive samplers and discrete data across a riverine system can provide a more comprehensive understanding of cyanoHABs over space and time.

The U.S. Environmental Protection Agency (EPA) issued drinking-water health-advisory values and recommendations for recreational water-quality criteria for microcystins and cylindrospermopsin (U.S. Environmental Protection Agency, 2019; table 1). National guidelines for anatoxins or saxitoxins do not currently exist, and many States have developed State-specific guidelines in response to the rise in occurrence of cyanoHABs using several different metrics and thresholds to define potential harm to recreational or water-supply users (Stackpoole and others, 2024). The New Jersey Department of Environmental Protection (NJDEP) has developed recreational State guidance levels for microcystin, anatoxins, and cylindrospermopsin, and has proposed health advisory guidance levels for these three cyanotoxins and saxitoxin in treated drinking water (New Jersey Department of Environmental Protection, 2022, 2024; table 1). For recreational contexts, NJDEP has developed four-tiered harmful algal bloom (HAB) Alert Levels: “Watch,” “Advisory,” “Warning,” and “Danger,” based on cyanobacteria cell concentrations and cyanotoxin concentrations (New Jersey Department of Environmental Protection, 2022; table 2). In New Jersey, the potential and number of documented cyanoHABs has increased in recent years (Hsu and others, 2024); therefore, an improved understanding of sources and transport of cyanoHABs within the State can help water-resource managers design efficient monitoring programs and allocate resources to protect human health.

From August 2020 to August 2021, the U.S. Geological Survey (USGS), in cooperation with the New Jersey Water Supply Authority (NJWSA), the NJDEP, and Montclair State University conducted a study to characterize the extent, duration, and potential transport of cyanobacteria and associated cyanotoxins within the Raritan Basin Water Supply Complex to better understand cyanotoxin and cyanobacteria occurrence in this system. The complex consists of Budd Lake, Spruce Run and Round Valley Reservoirs, and the main stem of the Raritan and Millstone Rivers (fig. 1). Collectively, the Raritan Basin Water Supply Complex provides drinking water to more than 1.5 million people in New Jersey (Hsu and others, 2024). The study included extensive discrete-sample collection, use of continuous water-quality sensors, and deployment of SPATT samplers throughout the year.

Previous analysis of phytoplankton data found several cyanobacteria genera, including 24 known cyanotoxin producers that were present in several sites throughout the supply complex and a net decrease in phytoplankton abundance from upstream to downstream sampling sites. It was also shown that persistence mechanisms differed between the Millstone and Raritan River subwatersheds (Clonan, 2026). Using multivariate methods, the authors showed that the lake and river sampling sites had substantially different cyanobacteria assemblages—all mainstem sites were similar, but Spruce Run Outlet and Budd Lake Outlet were each distinct. Using flow data collected at USGS streamgages, Clonan and others (2026) determined cyanobacteria persistence in the Raritan River overlapped with discharge from Spruce Run Reservoir; however, the study did not address cyanotoxin occurrence in the system, the use of genetic methods and passive samplers, or additional phytoplankton and chlorophyll a data collected from within Spruce Run Reservoir.

The Raritan River Basin comprises an area of 1,105 square miles (mi²) in New Jersey (fig. 1). It is the largest river drainage basin entirely within New Jersey and is affected by municipal and industrial point-source discharges (New Jersey Water Supply Authority, 2005). The basin drains all (or parts of) around 100 municipalities in seven counties. The confluence of the mainstem of the Raritan River with the Millstone River (fig. 1) is the location of two drinking-water intakes (New Jersey Water Supply Authority, 2005). Numerous low-flow dams modify river flow, and two wastewater-treatment plants discharge 8.5 mi upstream of the Blackwells Mills MR site and 2 mi upstream of the Below Calco Dam RR site, respectively, which can affect water quality (Clonan, 2026).

The Raritan River Water Supply Complex was selected for this study because it is a source of drinking-water supply and the recurrent cyanoHABs in upstream lakes and reservoirs (New Jersey Water Supply Authority, 2005; Hsu and others, 2024; Clonan, 2026). There are several lake-like portions within the basin, including Budd Lake and Spruce Run Reservoir (fig. 1). Budd Lake¹ 1

The surface area of Budd Lake equals (=) 376 acres (New Jersey Department of Environmental Protection, 2025).

This study used a combination of discrete water-quality and SPATT sampler data that were collected on 20 dates from 8 riverine and lake outlet locations in the Raritan River Water Supply Complex (table 3). Additionally, 24 discrete water-quality samples were collected at the Spruce Run Buoy site within Spruce Run Reservoir (fig. 1) to record continuous water-quality data (collected in 15-minute intervals).

Quality control measures are an important part of laboratory and field best practices. Approximately 20 percent of the discrete samples collected at each site included concurrent field replicates and field blanks for quality assurance and control purposes. Concurrent replicates were analyzed for most constituents. Concurrent replicates were collected over a range of streamflow conditions and at various times throughout the year. Laboratory and field blanks were used to identify potential sources of contamination related to equipment cleaning and sample-collection procedures.

In this study, there were no detections in field blanks for TN, but there were two dates with detections of TP. One was about half of the ambient TP (0.035 mg/L on May 11, 2021), and the other was much higher than the corresponding field value and flagged by the lab as “estimated” (0.596 mg/L on Oct 6, 2020). The latter detection was related to a contamination issue with the deionized water source used at that time and not an analytical error, so the data were retained. The relative percentage difference (RPD) was utilized to assess the variations in concentration between replicate water samples. The RPD was computed by taking the difference in concentration between two replicates for TN, TP, and chlorophyll a, dividing it by the mean concentration of the replicate pair, and then multiplying the result by 100. This calculation yields a value representing the percentage difference in concentration between the replicate samples (Zar, 1999). All RPDs between replicate pairs had median RPD values of less than 10 percent for field replicates; for TP: median RPD=6.9, n=20 replicate pairs; TN median RPD=2.3, n=20 replicate pairs; and chlorophyll a median RPD=7.7, n=19 replicate pairs. For laboratory replicates, chlorophyll a median RPD=3.6, n=20 replicate pairs.

For the phytoplankton samples analyzed by NJCWST, absolute value logarithmic difference (AVLD) was used to evaluate differences in cyanobacteria between replicate pairs (Francy and others, 2015). The AVLD was calculated as:AVLD=log10R1−log10R2,where

AVLD was used to evaluate differences for total phytoplankton data instead of RPD calculations because it is less sensitive to rare taxa present in one of the replicate samples, but not the other; replicate pairs with an AVLD value of less than 1.0 were considered acceptable for this study. Only 1 sample pair (from Stony Brook on October 20, 2020, AVLD=1.9) was above this threshold. The environmental sample was retained in this study for comparison to other samples in the context of the overall low abundance of phytoplankton at the site, and in this particular sample, an excess of a few rare taxa caused an elevated AVLD. Excluding that pair, the AVLDs for phytoplankton across all other replicate pairs ranged from less than 0.01 to 0.88 (median AVLD: 0.28, n=19 replicate pairs). All data were considered acceptable quality for the purposes of this report. Replicate samples were not collected at Spruce Run Buoy for analysis at PhycoTech Inc.

For the genetic analyses, all laboratory quality-control blanks (filtration blanks, extraction blanks, and qPCR no-template controls) were negative for all targets. All sample extracts, standards, and blanks were analyzed in duplicate. There were no detections in any lab blanks. For discrete samples analyzed for total microcystins by ELISA, quality-control measures (including percent coefficient of variation between two well replicates, coefficient of determination [R²], usage of laboratory reagent blanks, low-calibration verification, usage of laboratory-fortified blanks, laboratory-fortified sample matrices and duplicate, and quality-control samples) were utilized in accordance with EPA Method 546 to ensure the accuracy and precision of analyses and standard curves (Zaffiro and others, 2016).

As with many environmental studies, field replicates and blanks can be used to ensure data validity and help with SPATT sampler-data interpretation (Jaegge and others, 2026; U.S. Geological Survey, variously dated). Two blank SPATT samplers were activated, stored, extracted, and analyzed in the same manner as the deployed SPATT samplers, and analysis by LC-MS/MS did not detect any cyanotoxins in the extracts. For field replicates, coefficients of variation are an appropriate indication of the range in values in SPATTs (Johnston and others, 2025). The coefficients of variation for microcystin concentrations ranged from 0.4 to 86.9 ng/g/day, with the median=38.3 ng/g/day for 6 replicate pairs. Given the lack of standardized methods for evaluating SPATTs (Jaegge and others, 2026; U.S. Geological Survey, variously dated), all samples were retained for analysis regardless of observed variation in results. SPATT results presented in this report should be considered a semiquantitative indicator of cyanotoxin presence.

A total of 184 discrete water-quality samples were collected among eight sites, including quality-control samples such as replicates and blanks. Discrete water-quality data were collected over a variety of streamflow conditions at all of the sites, except for the highest 10 percent of flow conditions, as personnel safety would have been a concern (fig. 2). The mean streamflow over the study period was lowest at Spruce Run Outlet but streamflow values were consistently lowest at Stony Brook (table 4; fig. 2) and was not measured at Budd Lake Outlet, which is characterized as an impoundment and generally has low flow or nonflowing conditions.

Over the course of this study, the sites had a substantial range when compared to one another for several water-quality characteristics (table 4). Nutrients (TN and TP) were highest at the furthest downstream mainstem Raritan River and Millstone River sites (Below Calco Dam RR and Blackwells Mills MR, respectively). Spruce Run Outlet also had elevated TP, but not TN. Median chlorophyll a was about 2–8 times higher at lake outlet sites than the riverine locations and was about 2 times higher at Budd Lake Outlet than Spruce Run Outlet. When compared to other sites, turbidity was consistently highest at Budd Lake Outlet. Turbidity was highest at the lake outlet sites in the summer months—concurrent with the periods of highest phytoplankton detections. The two largest riverine sites (Manville RR and Below Calco Dam RR) had relatively elevated median turbidity values compared to the upstream riverine sites, but this was largely driven by a high-flow event around February 16, 2021.

In this study, there were substantial differences between lake outlet and riverine sites in overall phytoplankton assemblage, total taxa richness, and abundance of cyanobacteria (figs. 3, 4). The most abundant phytoplankton divisions (followed by percent abundance across all sites and dates) were: Cyanobacteria (91 percent), Chlorophyta (5 percent), and Bacillariophyta (3.3 percent). The most common cyanobacteria taxa (followed by percent abundance across all sites and dates) were: Cyanodictyon/Aphanocapsa (70.3 percent), Aphanathece (10.6 percent), Microcystis (3.9 percent), Aphanizomenon (3.8 percent), and Planktothrix (3 percent). Lake outlet sites had higher phytoplankton densities and were more likely to be dominated by potential cyanotoxin-producing and filamentous cyanobacteria taxa. For example, Aphanizomenon was generally the most dominant cyanobacteria taxa at Spruce Run Outlet and was present in every sample collected at that site; however, it was only identified in 6 percent of samples collected at Manville RR. Several common bloom-forming taxa (Microcystis, Merismopedia, Dolichospermum, Leptolyngya, Planktolyngbya, and Planktothrix) were only detected at the two lake sites. Several small, coccoid (round), and likely noncyanotoxin-producing taxa (such as Cyanodictyon/Aphanocapsa, Synechococcus, and Chroococus) were detected at all sites (Treviño and others, 2025). Budd Lake Outlet had the strongest seasonally driven patterns in cyanobacteria concentrations (fig. 4A), with a total abundance of more than 1,000,000 cells/mL in October 2020, which is more than 50 times higher than the lowest abundance of approximately 18,000 cells/mL measured in February 2021.

The total abundance of cyanobacteria cells in discrete samples exceeded the NJDEP HAB Alert Watch threshold (table 2) in 38 samples out of 160 samples (24 percent of all samples), collected at four sites (Spruce Run Outlet, Budd Lake Outlet, Stanton RR, and Stony Brook). The NJDEP HAB Alert Advisory threshold (table 2) was exceeded in 18 samples (11 percent of all samples) from only the two lake sites (Spruce Run Outlet and Budd Lake Outlet).

At Budd Lake Outlet, the Watch threshold was exceeded in 95 percent of the samples (fig. 4A), and at Spruce Run Outlet, it was exceeded 80 percent of the time for total cyanobacteria (fig. 4B). When only considering potentially toxin-producing cyanobacteria taxa, the exceedance frequencies decline to 50 percent at both Spruce Run Outlet and Budd Lake Outlet. All but four samples at Budd Lake Outlet exceeded the NJDEP HAB Alert Advisory threshold for total cyanobacteria (table 2; fig. 4A); however, most samples were dominated by genera Cyanodictyon and Aphanocapsa, which are not known to produce cyanotoxins. The only dates when the threshold was exceeded by cyanotoxin-producing taxa at Budd Lake Outlet were July 20, 2021, and August 3, 2021, when the samples included Microcystis, Dolichospermum, and other cyanobacteria taxa. At Spruce Run Outlet, two samples exceeded NJDEP HAB Alert Advisory threshold for total cyanobacteria, but neither were dominated by cyanotoxin-producing taxa (fig. 4B).

Exceedances of the NJDEP HAB Alert Watch threshold (table 2) occurred less than 3 percent of the time at the riverine sites (fig. 4C, D, E, F, G, and H). The only date when potentially cyanotoxin-producing taxa were above the threshold was July 20, 2021, at Stanton RR, when the sample was comprised in part of Fischerella (20,136 cells/mL), which was the only sample where that taxa was detected across the whole study period (Treviño and others, 2025). For two other dates when total cyanobacteria cell counts exceeded the NJDEP HAB Alert Watch threshold at riverine sites (May 11, 2021, at Stanton RR and February 16, 2021, at Stony Brook), the composition was dominated by taxa not known to produce cyanotoxins, such as Cyanodictyon, Aphanocapsa, and Aphanothece. No samples in this study exceeded the NJDEP HAB Warning or Danger thresholds (table 2).

At Spruce Run Buoy, samples are presented as biovolume (cubic micrometers per milliliter; µm³/mL) as analyzed by PhycoTech Inc. The most abundant phytoplankton divisions present across all samples on all dates (followed by total percent biovolume) were: Bacillariophyta (54.9 percent), Cyanobacteria (36.9 percent), Chlorophyta (5 percent), and Other (3.3 percent). The most common cyanobacteria taxa present across all samples on all dates (followed by total percent biovolume) were: Aphanizomenon (69.8 percent), Dolichospermum (15.8 percent), Limnoraphis (3 percent), Pseudanabaena (3 percent), Microcystis (2.9 percent), Chrysosporum (1.7 percent), and Woronichinia (1.3 percent). This is consistent with the phytoplankton composition detected at Spruce Run Outlet in the samples analyzed by NJCWST.

A comparison of the continuous-sensor fluorescence data for chlorophyll a, phycocyanin, and discrete chlorophyll a (fig. 5A) from the Spruce Run Buoy reveals some discrepancies with the trends observed in microscopy biovolume. The patterns observed in the sensor data for chlorophyll a and phycocyanin are not always in sync with the total phytoplankton biovolume (fig. 5B) or the cyanobacteria biovolume (fig. 5C) determined from the discrete samples. The chlorophyll a sensor indicated two relative peaks in algal biomass during the study period: from November 2020 to February 2021 and from April 2021 to May 2021 (fig. 5A). Both periods aligned with elevated discrete sample chlorophyll a concentrations, and phytoplankton growth in the April to May 2021 period; however, phytoplankton abundance in the discrete samples was not similarly elevated in the winter period. There was little variability in the phycocyanin data over the entire study period. Both methods indicated a steady rise in algal concentration from March to June 2021, followed by a steep decline beginning on May 25 and continuing to June 8, 2021. The phycocyanin sensor data were consistent throughout the study period and were less sensitive to seasonal patterns of increase or decrease, similar to the other parameters. These findings limit the potential of only using continuous data to determine the occurrence of blooms.

16S cyanobacteria genes were detected in all (100 percent) of the 59 analyzed samples from the subset of five sites (Budd Lake Outlet, High Bridge RR, Spruce Run Outlet, Stanton RR, and Manville RR; table 5; fig. 6A), which was higher than the frequency cyanobacteria was found by microscopy (87 percent) in the 97 phytoplankton samples collected from all 8 study sites. The highest abundance of gene copies was at Budd Lake Outlet on October 20, 2020, with 2.2×10⁸ copies per 100 mL. Overall, the 16S cyanobacteria rRNA was strongly positively associated with the abundance of total cyanobacteria (R²=0.62; p>0.001; fig. 7A). On sample dates when no cyanobacteria were detected by microscopy, the cyanobacteria 16S rRNA was still detected (fig. 7A); however, the presence of cyanobacteria 16S rRNA does not provide information regarding the specific taxa or potential for cyanotoxin occurrence.

All four classes of cyanotoxin synthetase genes (mcyE, anaC, sxtA, and cyrA) were detected at least once among all analyzed samples; however, the frequency of detection (percent occurrence) and abundance of gene copies varied substantially by site (table 5, fig. 6B, C, D, E) and over time (fig. 8B, C, D, E). For anaC, 77 percent of riverine samples had detections, whereas 66 percent of samples from lakes had detections. The synthetase gene mcyE was detected in all lake outlet site samples throughout the study. In contrast, 65 percent of riverine samples had detections, which is a lower rate than a previous national study of large rivers that found mcyE in about 90 percent of riverine samples (Zuellig and others, 2021). Compared to the riverine sites (table 3), the lake outlet sites (Spruce Run Outlet and Budd Lake Outlet) had higher median abundance of mcyE and sxtA (fig. 6B, D); but, for anaC, abundance at the riverine sites was higher (fig. 6C), which is consistent with previous studies (Howard and others, 2021; Zuellig and others, 2021). The highest single-sample abundance of sxtA was from Spruce Run Outlet, with 100,000 copies/100 mL detected on September 22, 2020; mcyE, Budd Lake Outlet, with 2,200,000 copies/100 mL detected on July 7, 2021; and cyrA, Budd Lake Outlet, with 5,200 copies/100 mL detected on June 8, 2021. The highest abundance of anaC was at Stanton RR, with 21,000 copies/100 mL on February 16, 2021.

The two lake outlet sites were dissimilar—Spruce Run Outlet had a higher rate of detection and higher mcyE gene concentrations throughout the study (table 5; fig. 6B, 8B). There were seasonal differences in phytoplankton composition, such as a shift to higher concentrations of microcystin-producing taxa. For example, Budd Lake Outlet had summer and fall peaks in mcyE (fig. 8B), which is consistent with seasonal patterns in total cyanobacteria abundance (fig. 4A). These patterns included a consistently high abundance of cyanobacteria at Budd Lake Outlet (fig. 4A), but a low persistence of the assemblage downstream to High Bridge RR (fig. 4C) which was also reflected in the mcyE results (figs. 6B, 8B). A different pattern was observed where the cyanobacteria assemblage from Spruce Run Outlet (fig. 4B) would often persist to Stanton RR (fig. 4D). This is mirrored in the mcyE results from Spruce Run Outlet to Stanton RR, and even further downstream to Manville RR on certain dates, such as May 11–12, 2021 (figs. 6B, 8B).

Changes in anaC presence can be linked to specific cyanobacteria taxa, and across all sites, the number of gene copies was significantly positively associated with total cyanobacteria density (fig. 4, fig. 7C). At Stanton RR, when there was a peak in anaC concentration on February 16, 2021 (fig. 8C), the cyanobacteria in that sample were comprised of: Aphanothece (40 percent abundance), Aphanizomenon (36.5 percent abundance), Dolichospermum (11.3 percent abundance), and Cyanodictyon/Aphanocapsa (7.8 percent abundance). That sample from Stanton RR was the only time Dolichospermum had appreciable abundance at the site throughout the study. It is noteworthy that the abundance of Dolichospermum at Stanton RR (1,007 cells/mL) on February 16, 2021, was about 2 times higher than at the Spruce Run Outlet on the same date (620 cells/mL; fig. 4) and indicates local growth may have been occurring despite the cold-water temperature on that date (3.5°C). The abundance of anaC at Spruce Run Outlet peaked on May 11, 2021 (fig. 8C), and the associated phytoplankton sample was composed of Jaaginema and Aphanizomenon (53.5 and 41.6 percent of total phytoplankton abundance, respectively), representing one of the lowest percent compositions by Aphanizomenon recorded at the site throughout the study. Aphanizomenon is a known anatoxins producer (Christensen and Khan, 2020), but Jaaginema is not known to produce any cyanotoxins. Cyanotoxin production is strain-specific (Wehr and others, 2014; Christensen and Khan, 2020). It is possible that the Aphanizomenon population had a higher proportion of strains with the anaC gene in May 2021. Alternatively, Jaaginema may have the anaC gene, although it has not yet been identified as a potential anatoxin producer. SxtA and cyrA were detected substantially less frequently than the other two synthetase genes (table 5; figs. 6D, E) and, when compared to the other two genes, were not significantly correlated to total cyanobacteria density (figs. 7D, E); therefore, there was no attempt to correspond increases in gene copies to the occurrence of specific cyanobacteria taxa.

Of the discrete samples analyzed for total and dissolved microcystins (177 analyzed samples), none had detections for dissolved microcystins above the method reporting limit (0.3 µg/L), although 15 samples were above the MDL (0.15 µg/L). For total microcystins, 11 samples (6 percent) had detections above the MRL, and an additional 20 samples were between the detection and reporting limits (table 6). The samples with detections above the reporting limit were exclusively from the 2 lake outlet sites (8 from Budd Lake Outlet and 3 from Spruce Run Outlet); this suggests that microcystins are present in the lakes, but they are not persisting downstream to the riverine sites. The highest single-sample microcystin detection was 1.4 µg/L on August 3, 2021, at Budd Lake Outlet, which was above the EPA-treated drinking-water thresholds (table 1), but is well below the threshold to determine a NJDEP HAB Alert Warning level (table 2) for recreational activities (New Jersey Department of Environmental Protection, 2022).

The SPATT samplers were retrieved after deployments ranging from 7 to 20 days, with a median value of 14 days across all samplers. Microcystin concentrations in SPATT extracts are semi-quantitative and should not be directly compared to the discrete sample results in water samples, as the SPATT results are representative of a span of time as opposed to an instantaneous measurement, and SPATT samplers were deployed for varying amounts of time (table 6). Anatoxins and cylindrospermopsins were not detected in any of the extracts, so the focus of this section is on the microcystin data only.

Microcystins were detected in 83 percent of SPATT samplers (n=60), which was a substantially higher detection rate than discrete samples analyzed for microcystin by ELISA (table 6), but similar to rates of detection of mcyE (table 5). The only sites with microcystin detections above the MRL in discrete samples (Budd Lake Outlet and Spruce Run Outlet) also had detections in SPATT samplers 100 percent of the time (fig. 9). Although the lake outlet sites had more frequent detections of microcystin, the two highest SPATT-sampler microcystin concentrations across all dates were from Stanton RR and Manville RR. It is important to note that they were from the first samplers deployed at those sites (August 26 to September 9, 2020) (fig. 10) and were not coincident with elevated concentrations of either potentially toxic cyanobacteria or mcyE gene copies at either site (figs. 4D, E, 8B). Though there was a lack of microcystin detections above the MRL in the discrete samples, especially from Stanton RR and Manville RR (table 6), the detections in the SPATT sampler results suggest that transport of microcystin downstream from the lake sites may have occurred. The third-highest single-sample value was from Budd Lake Outlet by the SPATT sampler deployed from June 22 to July 7, 2021, which was coincident with elevated levels of potentially toxic cyanobacteria and mcyE gene copies (figs. 4A and 8B). These samplers were all deployed during peak cyanoHAB season, which is generally July through September in New Jersey (New Jersey Department of Environmental Protection, 2022); SPATT cyanotoxin-accumulation rates were lowest during the winter months (October through April; fig. 10).

Evidence of movement of cyanobacteria and microcystin among sites and over time was evaluated by aligning four indicators: total phytoplankton abundance, detections of mcyE by qPCR, microcystin analysis of discrete samples, and microcystin analysis from SPATT samplers (fig. 11). Detections across all four indicators and agreement among the methods were most common at the two lake outlet sites (Budd Lake and Spruce Run; figs. 11A, C). There were no seasonal trends in detection that suggest downstream transport was more prevalent in the summer or winter months (fig. 11). Clonan (2026) showed that cyanobacteria cells did not often transport downstream in this system; however, this analysis shows that there was some transport of both genetic material and microcystin in sporadic detections by qPCR and SPATT samplers, respectively. The data from Budd Lake and Spruce Run Outlets suggest that these sites were a constant source of cyanobacteria, potential cyanotoxin producers, and cyanotoxins, with several dates where all four cyanoHAB indicators were detected (figs. 11A, C). However, the cyanoHABs did not appear to transport downstream. This was particularly apparent during the summer when cyanoHAB conditions occurred at the lake outlet sites, but were not detected at the riverine sites, such as the period from July 7, 2021, to August 3, 2021 (fig. 11). At the riverine sites, several dates were discordant, wherein cyanotoxins were detected by one indicator and not the others (figs. 11B, D, E). The cyanotoxins and genes that were detected at downstream sites (Stanton RR and Manville RR) were generally not linked to bloom conditions and may have been produced locally or transported from other source areas that were not sampled as part of this study.

Overall, the strength of significant Spearman-correlation associations between biological response and environmental variables (11 total variables, 55 correlations, fig. 12) was stronger in lake sites than riverine sites (median values=0.41 and 0.15, respectively) and more often positively correlated (87 percent and 56 percent, respectively). Relations were nearly all positive at the lake sites, except for several indicators (TN, anaC, and cyanobacteria 16S rRNA) with SPATT microcystin detections (fig. 12A), which is consistent with the restrictions of relying on SPATT as an environmentally relevant indicator of environmental concentrations (Howard and others, 2021, Jaegge and others, 2026). At the riverine sites, TN and TP were negatively correlated with most of the biological indicators (fig. 12B), which is consistent with other literature on riverine HABs (Wood and others, 2020; Jarvie and others, 2025). Further, several specific correlation pairs between the biological indicators were stronger in riverine sites despite less frequent detections, such as total abundance of cyanobacteria with chlorophyll a, cyanobacteria 16S rRNA, and mcyE gene copies (fig. 12B).

The widespread occurrence of cyanobacteria throughout the study area was consistent with genetic findings, suggesting a common presence of cyanobacteria in the Raritan Basin Water Supply Complex. Although discrete sampling during this study did not reveal cyanotoxins above any thresholds to human health (tables 1, 2), the presence of toxigenic cyanobacteria, cyanotoxin-producing genes, and consistent detections of microcystins from the SPATT samplers indicate potential for future concerns related to cyanoHABs within the Raritan Basin Water Supply Complex. The low cyanotoxin concentrations likely do not have human health implications, but they may affect aquatic and terrestrial food webs (Moy and others, 2016; Bukaveckas and others, 2018). When conditions are favorable, particularly during warm temperatures from June through September and (or) during periods of low flow, cyanobacteria could begin to proliferate and lead to more widespread cyanotoxin occurrences. Considering the Raritan Basin Water Supply Complex’s extensive use as a drinking-water supply, these cyanoHAB events could lead to potential risks to human and environmental health (Hsu and others, 2024). Cyanobacteria were typically most abundant in August and September, but cyanoHABs can occur any time of year; therefore, datasets that span year-round, such as this study, can enhance understanding of bloom-formation mechanisms (Chorus and Welker, 2021; Reinl and others, 2023).

These findings are consistent with previous studies, which indicated that cyanobacteria, cyanotoxin synthetase genes, and cyanotoxins have been found at low concentrations in many large rivers across the Nation, with eutrophic (high-nutrient) rivers having the highest phytoplankton biomass, abundance of cyanotoxin synthetase genes, and cyanotoxin occurrences (Graham and others, 2020; Linz and others, 2023). The sporadic detection of microcystins by ELISA compared to the persistent detection of both genes by qPCR (cyanotoxin synthetase genes and 16S rRNA) and microcystins by SPATT suggests there could be sources of cyanobacteria growth elsewhere in the system in addition to Budd Lake and Spruce Run Reservoir.

In this study, detections of cyanobacteria 16S rRNA and cyanotoxin synthetase genes had higher sensitivity to detect cyanobacteria compared to microscopy, which has also been shown in other studies for large rivers in New Jersey (Hsu and others, 2024) and across the United States (Graham and others, 2020). Cyanobacteria abundance and mcyE gene copies across all sites were moderately correlated and significantly positively associated (R²=0.55, p<0.001; fig. 7B). Relative mcyE densities between upstream and downstream sites on each sampling date overlapped with patterns in taxa capable of producing microcystin, as described by Clonan (2026).

Because of their continuous-absorption capacity, SPATT samplers are well-suited for the evaluation of spatial and vertical cyanotoxin dispersal and persistence patterns in flowing-water systems (Kudela, 2011; Wood and others, 2020; Howard and others, 2021). Other studies have advanced the understanding of SPATT use and applicability in a range of system types (Carpenter and Wise, 2023; Johnston and others, 2025; Jaegge and others, 2026). However, there are substantial limitations to extrapolating quantities of cyanotoxins detected in SPATT-sampler extracts to environmentally relevant concentrations. In addition to deployment time, cyanotoxin adsorption by SPATT samplers can be affected by water temperature, turbidity, water chemistry, salinity, hydrodynamics, and concentrations of cyanotoxins (Jaegge and others, 2026). Results from the SPATT deployments in this system show persistent low concentrations of microcystin compared to the discrete samples (figs. 9, 11). At times, cyanotoxins can be detected with SPATT in the absence of any other measures, which indicates they may be transported from upstream locations in low concentrations rather than cyanobacteria growth occurring at each site. SPATT samplers can serve as an early warning tool for source-water reservoirs or downstream-intake points to detect cyanotoxins prior to the occurrence of a bloom event (Howard and others, 2021).

During periods of increased streamflow, greater cyanobacteria biomass can be mobilized and transported downstream, potentially affecting water quality and ecological health. Graham and others (2012) support this notion, demonstrating a correlation between larger reservoir releases and increased cyanobacteria densities in downstream waters. These findings emphasize the importance of considering hydrological factors—such as streamflow and reservoir releases— when assessing the potential for cyanotoxin transport and its effects downstream (Schmadel and others, 2024; Clonan, 2026).

Although continuous water-quality sensors provide valuable data, the findings from the Spruce Run Buoy suggest that, at least at that site, further data collection and analysis could help to: (1) accurately interpret the algal growth trends, and (2) observe indicators of algal blooms in real time. This finding aligns with previous research that highlighted the limitations of fluorescence sensors when detecting cyanoHABs (Foster and others, 2022). This discrepancy may be attributed to changes in algal material following sample preservation (Gifford and others, 2024) as well as differing water-column depths between the sensor location on the buoy and the depth at which discrete water samples are collected. Additionally, the fluorescence sensor measures a biological response affected by factors such as light environment, physiological status of algal communities, and algal morphology (Foster and others, 2022). Potentially, continuous water-quality data collected concurrently can help identify environmental variables that may affect the adsorption of cyanotoxins to SPATT as well as natural materials in the lake and rivers.

This study represents a snapshot of conditions from August 2020 to August 2021. To fully understand the complex dynamics of cyanotoxin transport and the effect of environmental factors, long-term monitoring studies may better capture seasonal and interannual variations. Seasonal changes in temperature, nutrient availability, and hydrological conditions can substantially affect the growth, cyanotoxin occurrence, and transport of cyanobacteria. By examining seasonal and interannual patterns, scientists and managers can gain valuable insights into the factors that drive cyanotoxin occurrences and develop more effective strategies for mitigating their effects.

A primary limitation in the ability to assess cyanotoxin transport in this study was the low number of cyanotoxin detections, especially cyanotoxins other than microcystin. This could be attributed to several factors:1. Temporal and spatial sampling resolution: The sampling design aimed to capture potential cyanotoxin-transport events, but the frequency and location of sampling sites may not have been optimal for detecting sporadic or localized occurrences. The timing of the study captured portions of two years (2020 and 2021), but not the full pattern of formation and dissipation of one bloom season.2. Low cyanotoxin concentrations: Cyanotoxins may have been present in the system, but were below MDLs, particularly during periods of low cyanoHAB activity, such as the winter months. The findings of more common detections in the SPATT samplers support this result as they tend to concentrate even low amounts of cyanotoxins (Kudela, 2011).3. Rapid cyanotoxin degradation: Environmental factors such as sunlight, temperature, and microbial activity can rapidly degrade cyanotoxins, limiting their persistence in the water column and reducing the likelihood of detection, especially anatoxins (Christensen and Khan, 2020; Chorus and Welker, 2021). The discrete samples were not analyzed for anatoxins; however, anatoxins and cylindrospermopsin were measured but not detected in the SPATT samples. Although potential producers for these cyanotoxins were present in the system, microcystin is likely the most concerning cyanotoxin in the Raritan Basin Water Supply Complex.4. Complex hydrological and meteorological conditions: Variable hydrological conditions, such as fluctuating flow rates and water levels, can affect cyanotoxin transport and dispersion, making it difficult to predict and track their movement. The study does not fully encompass the scope of interannual variability. For example, during times of drought, discharge from Spruce Run Reservoir makes up a much higher proportion of the Raritan River flow than the portion of discharge the reservoir contributed to the river during this study (New Jersey Water Supply Authority, 2005).

Further research could increase understanding of the factors influencing cyanotoxin transport and support development of more effective monitoring strategies. Future studies could benefit from increased sampling frequency, the use of more sensitive analytical techniques, and the integration of hydrological and meteorological data to refine predictions of cyanotoxin-transport events. To further refine our understanding of cyanotoxin transport and its potential effect on drinking-water sources, future investigations could focus on the following key areas:

From August 2020 to August 2021, the U.S. Geological Survey (USGS), in cooperation with New Jersey Water Supply Authority (NJWSA), New Jersey Department of Environmental Protection (NJDEP), and Montclair State University conducted a study to characterize the extent, duration, and potential transport of cyanobacteria and associated cyanotoxins within the Raritan Basin Water Supply Complex to better understand cyanotoxin and cyanobacteria occurrence in this system. The specific objectives of this report are to: (1) characterize findings from discrete water-quality samples, solid phase adsorption toxin tracking (SPATT), and continuous-sensor measurements, and (2) relate those results to cyanobacteria, cyanobacteria genes, cyanotoxins, chlorophyll a, water-quality data, and relevant New Jersey cyanobacteria harmful algal bloom (cyanoHAB) advisory thresholds.

Data were collected on 20 dates from 8 riverine and lake outlet locations in the Raritan River Water Supply Complex. Discrete samples were analyzed for nutrients, turbidity, chlorophyll a, phytoplankton identification and enumeration, cyanobacteria genes, and cyanotoxins. A sonde was used to measure water temperature and pH at each site during discrete-sample collection. SPATT extract samples were analyzed for cyanotoxins.

This study showed substantial assemblage differences between lake outlet and riverine sites in overall phytoplankton composition. Lake outlet sites had higher densities of phytoplankton and were more likely to be dominated by potential cyanotoxin-producing and filamentous cyanobacteria taxa. On several dates, the abundance of cyanobacteria exceeded the NJDEP harmful algal bloom Alert Watch threshold (above 20,000 cells per milliliter [cells/mL]) at the lake sites. Comparatively few exceedances occurred at the riverine sites, which were rarely dominated by potential cyanotoxin-producing taxa. At the Spruce Run Buoy site, the phytoplankton composition was dominated by potential cyanotoxin-producing taxa throughout the year. Continuous-sensor-fluorescence chlorophyll a and phycocyanin data show some discordance with the microscopy biovolume, limiting the potential usage of continuous data to determine if a bloom is occurring using these data alone. All four classes of cyanotoxin synthetase genes (microcystin [mcyE], anatoxins [anaC], saxitoxin [sxtA], and cylindrospermopsin [cyrA]) were detected at least once among all analyzed samples; however, the frequency of detection (percent occurrence) and abundance of gene copies varied substantially by site and over time. Compared to the riverine sites, the lake outlet sites had higher mean detections of mcyE, sxtA, and cyrA, but for anaC, detections at the riverine sites were higher. Three of the genetic indicators (cyanobacteria 16S ribosomal ribonucleic acid [rRNA], mcyE, and anaC) were significantly positively associated with total cyanobacteria abundance.

Of the discrete samples analyzed for microcystins, none had detections above the method reporting limit (0.3 micrograms per liter [µg/L]). Microcystin detections were exclusively from the lake outlet sites and were all below the NJDEP HAB Alert Warning level [20 µg/L]). Microcystins were detected in 83 percent of SPATT samplers, a substantially higher rate than for discrete samples. Evidence of cyanobacteria and microcystin transport among sites was evaluated over time by aligning four indicators: total phytoplankton abundance, detections of mcyE by quantitative polymerase chain reaction (qPCR), microcystin analysis of discrete samples, and microcystin analysis from SPATT samplers. Detections across all indicators were most common at the two lake outlet sites. There were no seasonal trends in the indicators that suggested downstream transport was more prevalent in the summer or winter months. Under the conditions sampled during this study, results suggest the cyanobacteria and cyanotoxins from Budd Lake and Spruce Run Reservoir did not often lead to pronounced cyanobacteria and cyanotoxin presence at Manville RR or further downstream. There was some transport of both genetic material and microcystin shown in sporadic detections by qPCR and SPATT samplers, respectively, but no evidence of proliferation exists at these sites. The results from Budd Lake and Spruce Run Outlets suggested that these sites were a constant source of cyanobacteria, even when a visible cyanoHAB was not present. The strength of associations between biological response and environmental variables were stronger in lake sites than in riverine sites.

The widespread occurrence of cyanobacteria throughout the study area is consistent with the genetic findings, suggesting a common presence of cyanobacteria in the Raritan Basin Water Supply Complex. Although discrete sampling during this study did not reveal cyanotoxins at concentrations of high risk to human health, the presence of toxigenic cyanobacteria, cyanotoxin-producing genes, and consistent detections of microcystins from the SPATT samplers indicate potential for future concerns related to cyanoHABs within the system. Although continuous water-quality sensors provide valuable data, the findings from the Spruce Run Buoy suggest, at least at that site, further analysis could help to: (1) accurately interpret the algal growth trends, and (2) observe indicators of algal blooms in real time. Additional studies of cyanobacteria and cyanoHABs in the Raritan River basin could benefit from increased sampling frequencies, more sensitive analytical techniques, and the integration of hydrological and meteorological data to refine our understanding of cyanotoxin occurrence in riverine systems.