Data Package Metadata   View Summary

Visual Index Data

These are the methods used for mc_vis_index_wq.csv.

Most monitoring surveys that collect data on water quality and fisheries in the Delta also collect visual observations of Microcystis and other visually detectable algal blooms. Because Microcystis colonies are relatively easy to identify visually in the field, this visual ranking gives a general idea of when and where the most common harmful cyanobacteria in the Delta occur

A surface water sample is brought on board a research vessel in a bucket and the Microcystis concentration is ranked on a scale of 1–5, 1 meaning “absent” and 5 meaning “very high” (Flynn et al. 2022) (supplemental PDF figure 1). Although this method is imprecise, it is generally reliable on the whole for detecting Microcystis and giving a rough estimate of magnitude

Visual assessment data for this report come from five surveys. Data sets were also subset to only include observations in the Legal Delta, categorized into regions shown in the supplemental PDF (Figure 2).

Data sources:

• The Environmental Monitoring Program (EMP) is conducted jointly by DWR, the California Department of Fish and Wildlife (CDFW), and Reclamation and collects water quality, phytoplankton, zooplankton, and benthic invertebrate data throughout the Delta, Suisun Bay, and San Pablo Bay. The EMP has recorded Microcystis observations at each of its discrete stations since fall 2015, using the scale shown in Figure 2 2. The EMP also collects data on phytoplankton community composition via microscopic enumeration of grab samples, allowing an evaluation of which species are contributing to phytoplankton blooms. These data are collected at 24 fixed stations and up to four floating stations each month throughout the year (IEP 2020). These data are published annually on the Environmental Data Initiative repository.

• The CDFW Summer Townet (STN) Survey samples fixed locations from eastern San Pablo Bay to Rio Vista on the Sacramento River, and to Stockton on the San Joaquin River and a single station in the lower Napa River. The STN survey runs twice per month during June, July, and August and samples at 40 stations (Figure 2 2). The survey primarily monitors young-of-the-year fishes, but also measures zooplankton and environmental variables including water temperature (°C), water clarity (Secchi depth and nephelometric turbidity units [NTU]), and specific conductance (microSiemens per centimeter [µS/cm]). Visual observations of Microcystis have been collected since 2007. STN data are available via the CDFW website.

https://filelib.wildlife.ca.gov/Public/TownetFallMidwaterTrawl/TNS%20MS%20Access%20Data/TNS%20data/

• The CDFW Fall Midwater Trawl (FMWT) survey samples at fixed locations from eastern San Pablo Bay to the Cache Slough complex and Sacramento Deep Water Ship Channel, on the Sacramento River, and to Stockton on the San Joaquin River. This survey runs once per month during September, October, and November at 122 stations (Figure 2 2). The FMWT survey primarily monitors young-of-the-year fishes, but also measures zooplankton and environmental variables including water temperature (°C), water clarity (Secchi depth and NTU), and specific conductance (µS/cm). Visual observations of Microcystis have been collected since 2007. FMWT data are available via the CDFW website. https://filelib.wildlife.ca.gov/Public/TownetFallMidwaterTrawl/FMWT%20Data/

• DWR’s North Central Region Office (NCRO) conducts water quality and cyanoHAB sampling at stations throughout the South Delta (Figure 2 2). These samples include chlorophyll, nutrients, bromide, and organic carbon. When water samples are collected, the study also measures environmental variables including water temperature (°C), water clarity (Secchi depth and NTU), specific conductance (µS/cm), and a visual Microcystis index. NCRO data are available from DWR’s Water Data Library platform. https://wdl.water.ca.gov/waterdatalibrary/Map.aspx

Nutrient Data

These are the methods used to create the disc_nutr_chla.csv file.

Discrete nutrient (ammonium, nitrate + nitrite, and orthophosphate) and chlorophyll-a data were collected from four sources:

• The EMP collects discrete water quality grab samples at all stations where samples for phytoplankton community composition are collected. Water is collected using a flow-through system in which it is pumped into the shipboard laboratory either from a fixed intake 1 meter below the water’s surface, or from a Van Dorn water sampler, or via a submersible pump (IEP 2020). DWR’s Bryte Laboratory performed analyses for dissolved ammonium, dissolved nitrate + nitrite (hereafter referred to as “nitrate”), total Kjeldahl nitrogen, total phosphorus, dissolved orthophosphate, and chlorophyll-a, using EPA methods, American Public Health Association Standard Methods, or DWR-approved modifications of these methods (IEP 2020).

• The DWR NCRO collects discrete nutrient and chlorophyll-a data at six locations in the Central Delta surrounding Franks Tract. Chlorophyll-a samples were collected routinely from 2014 through 2021, while nutrient samples were collected only in 2014–2016 and 2021. Water is collected from a Van Dorn water sampler at a depth of 1 meter (DWR 2022). DWR’s Bryte Laboratory analyzed the samples using EPA methods or DWR-approved modifications of these methods (IEP 2020).

• The U.S. Geological Survey (USGS) has two programs that routinely collect discrete nutrient and chlorophyll-a data in the Delta: the California Water Science Center (CAWSC) and the San Francisco Bay Water Quality Survey (SFBS). The CAWSC collects samples at numerous locations throughout the Delta; the SFBS collects most of its samples downstream of the Delta, with a few locations extending into the Delta. The SFBS has been collecting discrete water quality samples from 1969 to present, while the CAWSC began collecting samples more recently.

• Data collected in 2014–2021 from the four surveys listed above were acquired through direct data requests or downloaded from either the discretewq data package (Bashevkin 2022), DWR’s Water Data Library, or the National Water Quality Monitoring Council’s Water Quality Portal. Data were integrated into one data set, limiting the stations to only those where all three nutrient parameters (ammonium, nitrate, and orthophosphate) and chlorophyll-a were collected. Some of the data collected in 2021 were considered provisional at the time of acquisition.

Outliers were identified as any value with a modified Z-score greater than 15 when the data were grouped by region. All identified nitrate and orthophosphate outliers were excluded from the data set. The detected ammonia and chlorophyll-a outliers were not removed because they appeared to be representative based on best professional judgment. Nutrient values that were below the reporting limit but had high reporting limits compared to the range of the overall data (greater than the 75th quantile) were excluded from the data set. In addition, the most common reporting limit for the laboratory method was used to estimate the reporting limit values for the nutrient data with missing reporting limit values. Additional details on data integration and processing can be found in the EDBdata GitHub package: https://github.com/mountaindboz/EDBdata.

Community Composition

These are the methods used to create the phyto_hab.csv file

Phytoplankton community composition data was obtained by microscopy from samples collected by EMP. These data were collected at 24 fixed stations and two stations that track the location of the salinity field each month throughout the year (see supplemental map, figure 2). Phytoplankton samples were collected with a submersible pump from a water depth of 1 meter below the water surface. Samples were stored in 50-milliliter (mL) glass bottles with 2 mL of Lugol’s iodine solution to act as a stain and preservative. Samples were analyzed by BSA Environmental Services, Inc. (Beachwood, Ohio). Phytoplankton were identified to the lowest taxonomic level possible, using the Utermöhl method and American Public Health Association Standard Method 10200 F (Utermöhl 1958, APHA 2017).

These data were subset to show only cyanoHABs species, defined as species in the genera Anabaeopsis, Aphanizomenon, Cylindrospermopsis, Dolichospermum, Oscillatoria, Planktothrix, and Microcystis. Although Microcystis is occasionally collected by these grab samples at a depth of 1 meter, particularly when the water column is well-mixed, it is better assessed by surface tows. These data are included to provide an idea of which taxa were present in the community, but the data should not be taken as a quantitative assessment of Microcystis abundance.

Satellite Imagery

This was used to create the hab_sat_ow_delta.csv file.

Satellite data, available from the San Francisco Estuary Institute’s HAB Satellite Analysis Tool (San Francisco Estuary Institute 2021), provides estimates of cyanobacterial. Satellite imagery is collected by the Ocean Land Color Instrument on the Copernicus Sentinel-3 mission. The cyanobacterial index algorithm (Wynne et al. 2018) is applied to the Ocean Land Color Instrument data to estimate cyanobacterial abundance in the upper portion of the water column by analyzing wavelengths of light that interact strongly with chlorophyll-a and phycocyanin, an accessory pigment in photosynthesis specific to cyanobacteria. Estimates of cyanobacteria abundance are reported in an exponential, satellite-specific, unitless metric called the Cyanobacteria Index (CI) for pixels with dimensions of 300 meters by 300 meters, each an area of approximately 22 acres.

Satellite mosaics of rasterized CI data across the Central Delta for June–October in 2020 and 2021 were downloaded from the San Francisco Estuary Institute’s HAB Satellite Analysis Tool (San Francisco Estuary Institute 2021). Raster pixels for four open-water regions in the Delta (Franks Tract, Clifton Court Forebay, Liberty Island, and Mildred Island) were extracted from each file using the ‘exact_extract’ function in the ‘exactextractr’ R package, version 0.7.1 (Baston 2021). The four open-water regions were defined using polygons derived from CDFW’s shapefile of Delta waterways and expanded by 200 meters around their perimeters to account for the large raster pixels.

Pixels were categorized into four CI categories (Low, Moderate, High, and Very High) based on WHO’s recreational guidance level thresholds (WHO 2021). Additionally, pixels that were below the detection limit for the imagery processing method (CI ≤ 6.310 x 10-05) were categorized as “Non Detect,” and pixels that were either invalid or missing were categorized as such. Including only pixels that were completely within one of the polygons of the four regions, the numbers of pixels within the “Non Detect,” “Invalid,” and four CI categories were counted for each region and raster image. Using only days when there were greater than 25 percent valid pixels within a region, the time series of pixel counts were visualized using area plots for each region and year.

Cyanotoxin data

This was used to create 'hab_toxins.csv'

The cyanotoxin data collected in 2021 and presented here came from six different sources (Figure 3, Table 1 in supplemental figures). Some of these sources had data available from previous years, but the majority of the data was from 2021, so only 2021 data are presented here, with the exception of Big Break, which contains data from 2015-2021. These studies all used either enzyme-linked immunosorbent assay (ELISA), liquid chromatography–mass spectrometry (LC-MS), liquid chromatography with tandem mass spectrometry (LC-MS/MS), or Abraxis test strips to analyze toxin concentrations. There is generally very high agreement between these two methods, although ELISA may produce higher concentration values than LC-MS/MS (Preece et al. 2021). Across most of the national harmful algal bloom (HAB) research community, data from either method are compared to thresholds, and no conversion factor is applied, nor is one method disregarded. Locations of samples are shown in supplemental pdf Figure 3 and Table 1.

• DWR collects cyanotoxin samples at Clifton Court Forebay and the Harvey O. Banks Pumping Plant (Banks Pumping Plant) to ensure that the water exported from the Delta is safe for use. Samples are collected every two weeks in April–October and analyzed by GreenWater Laboratories (Palatka, Florida), using a tiered approach. Samples are first assessed via microscopy to identify whether potentially toxic algae or cyanobacteria are present. If potentially toxic algae are detected, cells are lysed and samples are then tested for probable toxins using either ADDA-ELISA or LC-MS/MS, as appropriate (Foss and Aubel 2015).

• A special study was conducted collaboratively by USGS and DWR with funding from the Delta Regional Monitoring Program. Samples were collected at several stations throughout the Delta: Jersey Point (JPT), Decker (DEC), Middle River (MDM), Liberty Island (LIB), Rough and Ready Island (P8, DWR-EMP), and Vernalis (C10, DWR-EMP). For these efforts, cyanotoxins were measured in whole water discrete samples and using Solid Phase Adsorption Toxin Tracking (SPATT) samplers every two to four weeks. SPATTs are synthetic resin plates deployed in the water for an extended time to determine whether toxins are present over the entire time period. All (100 percent) of these cyanotoxin samples were to be analyzed using LC-MS/MS, and—upon review of LC-MS/MS data—a subset (approximately 20 percent) would be selected for analysis using ELISA. All laboratory analyses were conducted by Lumigen Instruments, Wayne State University, Detroit, Michigan. Data from this study have not been approved by USGS and are considered preliminary.

• The California State Water Resources Control Board’s freshwater HAB program collects samples for cyanotoxins when large blooms are reported (CCHAB Network 2022a). A special study conducted Ellen Preece, (Robertson Bryan Inc), Tim Otten (Bend Genetics LLC) and Janis Cooke (Central Valley Regional Water Quality Control Board; CVRWQCB) included collection of water samples for cyanotoxin analyses. This project was supported by the Delta Regional Monitoring Program and the State Water Resources Control Board Freshwater Harmful Algal Bloom Program. The Central Valley RWQCB collected cyanotoxin samples A single subsurface water grab sample was collected from the eastern side of Franks Tract on July 2 and again on August 6, 2021 (Preece et al, in preparation). Samples were lysed and analyzed by Bend Genetics, LLC (Sacramento, California) for total microcystins/nodularins, using the ADDA ELISA method.

• A special study of bioaccumulation of cyanotoxins in invertebrates in the Delta was conducted by David Senn (San Francisco Estuary Institute), Janis Cooke (Central Valley Regional Water Quality Control BoardRWQCB), Ellen Preece (Robertson-Bryan, Inc.), and Timothy Otten (Bend Genetics) (Preece et al, in preparation). This work was funded by a State of California Proposition 1 grant administered by the Department of Fish and Wildlife Ecosystem Restoration Program. Whole-water samples were collected monthly in the winter and every two weeks during the summer and analyzed for microcystins by Bend Genetics using ADDA ELISA.

• The East Bay Regional Park District (East Bay Regional Parks) conducts sampling at Big Break Regional Shoreline, visually inspecting the water for signs of cyanobacteria twice per month. If signs of cyanobacteria are detected, microscopy and toxin analysis are conducted at Bend Genetics using ADDA ELISA. Staff at East Bay Regional Parks requested data. Because Big Break has a longer monitoring history than most of these programs, all data for 2015–2021 were requested to get a better sense of how droughts and drought actions affect this cyanoHAB “hot spot.”

• Nautilus Data Technologies is required to monitor for cyanotoxins near its data center at the Port of Stockton. Nautilus Data Technologies monitors at six sites on the San Joaquin River and in the Stockton Deep Water Ship channel twice per month. All water samples are sent to Bend Genetics, where the samples are analyzed for microcystins, anatoxins and saxitoxins using ADDA ELISA as appropriate. Data were requested from staff at the State Water Board’s cyanoHABs portal.

None of the sources of cyanotoxin data presented here are part of a comprehensive monitoring program.

• The USGS/DWR SPATT study and the Proposition 1 Senn/Preece/ Cooke/Otten studies were designed as special studies to better understand toxin dynamics, rather than to establish a baseline. The RWQCB data are designed as a response to severe blooms, not a comprehensive monitoring program.

• The DWR Banks Pumping Plant/Clifton Court Forebay monitoring is designed specifically to assess water quality for water export, so it is not necessarily applicable to the rest of the Delta.

• Nautilus data are limited to the San Joaquin River, so they are unlikely to be influenced by the TUCP.

Combining these data sets does provide a relatively wide spatial and temporal scope of cyanotoxin monitoring, although it may miss small-scale or short-lived toxin events, particularly in smaller, backwater sloughs in the Delta. Different labs and field collection crews may result in slight biases in the resulting data sets, but all these data can be compared to the health advisory levels in the same way.

Fluoroprobe Data

This was used to create the Fluoroprobe.csv file

The EMP and USGS both employ vessels equipped with high-resolution sensors that collect data continuously on both water quality and phytoplankton community composition while underway. Both surveys monitor the phytoplankton community’s composition using a FluoroProbe instrument (bbe moldaenke GmbH, Schwentinental, Germany) that differentiates between cyanobacteria, diatoms, green algae, and cryptophytes, based on the wavelength of the fluorescence given off by each taxonomic group’s characteristic photopigments. USGS conducted mapping surveys in May, July, and October 2021, while EMP surveys are collected monthly throughout the year. Each month, these agencies covered approximately 350 miles of channels in the Delta over three to four consecutive days. USGS boat-based survey data can be visualized using USGS’s online data portal (https://tableau.usgs.gov/views/SFBD_Data_Portal/Mapping2018and2020?%3Aembed=y&%3AisGuestRedirectFromVizportal=y). FluoroProbe data collected by both the EMP and USGS were processed following the methodology described in the Methods PDF of the USGS data (Bergamaschi et al. 2020).

Continuous Water Quality

This was used to create 'cont_wq_daily_avg.csv', 'air_temp_daily_dd.csv' and 'water_temp_daily_dd.csv'

DWR and USGS maintain a network of water quality sondes and flow stations that collect data continuously (i.e., every 15 minutes) across the Delta (locations in supplemental pdf Figure 4). These stations collect data on water temperature, specific conductance, flow, DO, chlorophyll fluorescence, turbidity, and pH (although not all stations contain all sensors; see Table 2 in supplemental metadata file). Quality-controlled data were requested from DWR personnel when available, and provisional data were queried from the California Data Exchange Center (CDEC) if no finalized data were available.

To see how extended periods of high temperatures may drive Microcystis blooms, the number of degree-days over 19°C was calculated by averaging the daily maximum and minimum water temperature at seven stations in the South Delta. Degree-days are frequently used as a way to characterize total ambient thermal energy over the course of the year, and degree-days is considered a better measure of temperature regime than average temperature when predicting cyanobacterial abundance (Dupuis and Hann 2009, Pick 2016, Larson et al. 2018). This was converted to degree-days using the formula:

Degree-days = (Daily Max Temp – Daily Min Temp)/2 – 19

The same analysis was then conducted on air temperature, to see whether air temperature patterns were similar to water temperature patterns. Air temperature was not available for most stations in the Delta, but the nearest stations to the study region were chosen (Figure 4, Table 2 in supplemental information).

Hyperspectral Imagery Ground-Truthing

Around the time that hyperspectral imagery is collected each year, the CSTARS staff collects ground-truthing field data on the community composition of aquatic vegetation across the Delta, including areas in and around Franks Tract and Big Break. They have not sampled at Clifton Court Forebay because access to that area is restricted. Efforts are ongoing to clean and integrate the SAV data from this time series, but the authors of this report were able to acquire and present the data for 2021.

In 2021, this field survey took place from late July to mid-August. In Franks Tract (supplemental pdf Figure 6) and Big Break (Figure 7), the CSTARS staff sampled for SAV at 47 sites and 30 sites, respectively. A thatch rake with 34 teeth tethered with a long rope was used to collect rake samples. The rake was thrown off the side of the boat into the patch and allowed to sink to the bottom of the water column and then dragged in and pulled up. They recorded all species collected on the rake, as well as the percentage of the sample volume each species represented, to the nearest 10 percent.

Hyperspectral Imagery

Since 2004, hyperspectral airborne imagery has been collected by fixed-wing aircraft over the Delta in many years, although the time of year and spatial extent of these surveys have varied. Franks Tract has been included in all surveyed years (2004–2008, 2014–2021, see Figure 5 in Supplemental Figures file).

Survey methods for the hyperspectral imagery have varied somewhat among years, but the approach generally proceeds as described here for the 2018 survey. During this survey, HyVista Corporation (Sydney, Australia) used the HyMap sensor (126 bands: 450–2,500 nanometers, bandwidth: 10–15 nanometers) to collect imagery at a resolution of 1.7 meters by 1.7 meters. A diverse suite of inputs was derived from these images to capture reflectance properties across different regions of the electromagnetic spectrum, which track biophysiological characteristics useful for distinguishing types of plants. These intermediate inputs were generated using IDL scripts (IDL 8.01, ITT Visual Information Solutions) in ENVI (ENVI 4.8, ITT Visual Information Solutions).

Concurrent with imagery collection, ground-truthing surveys were conducted to determine species composition at points across the Delta region (e.g., 2018: 950 points; see the “Hyperspectral Imagery Ground-Truthing ” section for details). Field data were divided into training and validation subsets for image classification and independent validation of class maps. Training and validation polygons were overlaid on the raster images with generated inputs, and corresponding pixels within the raster images were extracted using the R statistical computing language (Version 4.0.2; R Core Team 2021) and packages ‘sp’ (Version 1.4.5) (Pebesma and Bivand 2021), ‘rgdal’ (version 0.5.5) (Bivand et al. 2021), and ‘rgeos’ (Version 1.5.23).

Training data were fed into a Random Forests classifier (packages ‘raster’: Version 3.4.5 (Hijmans 2021) and ‘randomforest’: Version 4.6.14 (Breiman 2001). The best-fit class type (e.g., open water, SAV, water hyacinth, water primrose) for each pixel was chosen based on consistency across tree predictions. The accuracy of the final maps was assessed using confusion matrices and Kappa coefficients. The area of SAV was calculated per year, per site, as the number of pixels classified as SAV multiplied by the area of a single pixel. FAV area was calculated in the same way, except that it is a combined category that includes water hyacinth, water primrose, and a mixed class composed of water primrose and emergent vegetation.

SePRO Vegetation Survey

DBW collaborates with SePRO Corporation to manage SAV in Franks Tract using the herbicide fluridone (Caudill et al. 2019). SePRO monitors changes in SAV community composition using point-intercept surveys (Madsen and Wersal 2018) that are conducted on one date annually in the fall.

Sampling points are chosen by generating a grid of evenly spaced points projected over the full area of Franks Tract (Figure 8, supplemental figures). The number of sampling points varies among years but is usually 100 (range: 50–200 samples). Most surveys have been conducted in mid October (range: October 1–October 13).

To sample each point, SePRO uses a weighted, double-headed, 0.33-meter-wide thatch rake attached to a rope, which is dragged for approximately 3 meters along the bottom and then pulled up to the boat for analysis. All SAV present on the rake is identified to species, and species-specific abundances are estimated based on the percentage of the rake each covers. Abundances are recorded using ordinal scores. During 2014-2018, scores ranged 1-4 (1 = 1–24 percent, 2 = 25–49 percent, 3 = 50–74 percent, 4 = 75–100 percent) and from 2019 onward, scores ranged 1-5 (1 = 1–19 percent, 2 = 20–39 percent, 3 = 40–59 percent, 4 = 60–79 percent, 5 = 80–100 percent).

Environmental drivers for weed data

Aquatic weed data were compared with water quality, flow, velocity, and herbicide application data to determine the drivers of variation in the abundance and composition of aquatic weeds. Variables hypothesized to affect aquatic weeds included measures of flow, turbidity, salinity, temperature, and herbicide applications (Figure 3-1). The analyses also included DO and pH, variables that are hypothesized to be affected by aquatic weeds.

Net Delta outflow data were obtained from DWR’s Dayflow model (DWR 2002). For water quality, monthly data were obtained from DWR’s EMP Station D19 (Franks Tract) and DFW’s Bay Study Station 853 (San Joaquin River just west of Big Break). The data for EMP Station C9 (Clifton Court) did not begin until recently (2016), so environmental drivers for this reference site were not considered. Discrete water quality stations were chosen over continuous stations for Franks Tract and Big Break because the discrete stations covered most parameters of interest for all years of aquatic vegetation monitoring (hyperspectral imagery started in 2004), whereas most continuous stations did not. As an exception, Bay Study Station 853 does not include DO or pH. For flow and water quality variables, annual means based on the main growing season for aquatic weeds (March–October) were used. Herbicide application data for Franks Tract (Table 3-1) and Clifton Court Forebay (Table 3-2) were obtained from DBW and DWR, respectively. The authors of this report are not aware of site-wide herbicide treatments in Big Break.

References cited in methods

American Public Health Association (APHA). 2017. Standard methods for the examination of water and wastewater. 23 edition. American Public Health Association, American Water Works Association, and Water Environment Federation,

Bashevkin, S. M. 2022. Six decades (1959-2020) of water quality in the upper San Francisco Estuary: an integrated database of 11 discrete monitoring surveys in the Sacramento San Joaquin Delta, Suisun Bay, and Suisun Marsh ver 2. Environmental Data Initiative. . https://doi.org/10.6073/pasta/dc199c9fd9c452337f01bcd9f0a89f50

Baston, D. 2021. exactextractr: Fast Extraction from Raster Datasets using Polygons. Comprehensive R Archive Network (CRAN), https://cran.r-project.org/web/packages/exactextractr/index.html

Bergamaschi, B. A., T. E. Kraus, B. D. Downing, J. Soto Perez, K. O'Donnell, J. A. Hansen, A. M. Hansen, A. D. Gelber, and E. B. Stumpner. 2020. Assessing spatial variability of nutrients and related water quality constituents in the California Sacramento-San Joaquin Delta at the landscape scale: High resolution mapping surveys: U.S. Geological Survey data release, . https://doi.org/10.5066/P9FQEUAL

Bivand, R., T. Keitt, and B. Rowlingson. 2021. Package 'rgdal': Bindings for the 'Geospatial' Data Abstraction Library. Comprehensive R Archive Network (CRAN), http://rgdal.r-forge.r-project.org/

Breiman, L. 2001. Random Forests. Machine Learning 45:5-32. 10.1023/A:1010933404324

Caudill, J., A. R. Jones, L. Anderson, J. D. Madsen, P. Gilbert, S. Shuler, and M. A. Heilman. 2019. Aquatic plant community restoration following the long-term management of invasive Egeria densa with fluridone treatments. Management of Biological Invasions 10:473-485. https://doi.org/10.3391/mbi.2019.10.3.05

DWR. 2022. Quality Assurance Project Plan: Central Delta and Emergency Drought Barrier Water Quality Monitoring Program. California Department of Water Resources, West Sacramento, CA,

Flynn, T., P. Lehman, S. Lesmeister, and S. Waller. 2022. A Visual Scale for Microcystis Bloom Severity. figshare Figure. https://doi.org/10.6084/m9.figshare.19239882.v1

Foss, A. J., and M. T. Aubel. 2015. Using the MMPB technique to confirm microcystin concentrations in water measured by ELISA and HPLC (UV, MS, MS/MS). Toxicon 104:91-101. https://doi.org/10.1016/j.toxicon.2015.07.332

Hijmans, R. J. 2021. Package 'raster': Geographic Data Analysis and Modeling. Comprehensive R Archive Network (CRAN), https://rspatial.org/raster

Interagency Ecological Program (IEP), S. Lesmeister, and M. Martinez. 2020. Interagency Ecological Program: Discrete water quality monitoring in the Sacramento-San Joaquin Bay-Delta, collected by the Environmental Monitoring Program, 2000-2018. ver 2. . Environmental Data Initiative. https://doi.org/10.6073/pasta/a215752cb9ac47f9ed9bb0fdb7fc7c19

Khanna, S., S. L. Ustin, E. L. Hestir, M. J. Santos, and M. Andrew. 2022. The Sacramento-San Joaquin Delta genus and community level classification maps derived from airborne spectroscopy data. Knowledge Network for Biocomplexity. http://doi.org/10.5063/F1K9360F

Madsen, J. D., and R. M. Wersal. 2018. Proper survey methods for research of aquatic plant ecology and management. Journal of aquatic plant management 56:90-96,

Pebesma, E., and R. Bivand. 2021. Package 'sp': Classes and Methods for Spatial Data. Version 1.4-6. Comprehensive R Archive Network (CRAN), https://github.com/edzer/sp/ https://edzer.github.io/sp/

Preece, E. P., W. Hobbs, F. J. Hardy, L. O’Garro, E. Frame, and F. Sweeney. 2021. Prevalence and persistence of microcystin in shoreline lake sediments and porewater, and associated potential for human health risk. Chemosphere 272:129581. https://doi.org/10.1016/j.chemosphere.2021.129581

Utermöhl, H. 1958. Methods of collecting plankton for various purposes are discussed. SIL Communications, 1953-1996 9:1-38. 10.1080/05384680.1958.11904091

World Health, O. 2021. WHO guidelines on recreational water quality: volume 1: coastal and fresh waters. World Health Organization, Geneva, https://apps.who.int/iris/handle/10665/342625

Wynne, T., A. Meredith, T. Briggs, W. Litaker, and M. p. d. R. Stumpf 2018. Silver Spring. 2018. Harmful Algal Bloom Forecasting Branch Ocean Color Satellite Imagery Processing Guidelines. NOAA Technical Memorandum NOS NCCOS 252. National Oceanic and Atmospheric Administration, Silver Spring, MD. 10.25923/twc0-f025