Loading...
HomeMy WebLinkAboutDWQ-2024-004547 1 Utah Lake Water Quality Study (ULWQS) Science Panel May 18, 9:00 AM to 1:30 PM Provo Airport – Skyview Lounge Meeting Summary - FINAL ATTENDANCE: Science Panel Members: Janice Brahney, Mike Brett, Greg Carling, Mitch Hogsett, Theron Miller, Hans Paerl, Thad Scott, and Tim Wool Steering Committee Members and Alternates: Eric Ellis Members of the Public: Jeff DenBleyker, Dan Potts, and David Richards Utah Division of Water Quality (DWQ) staff: Scott Daly and John Mackey Technical Consultants: Kateri Salk Facilitation Team: Heather Bergman and Samuel Wallace ACTION ITEMS Who Action Item Due Date Date Completed Tetra Tech Create a time series graph for different water quality variables across sites to explore the influence of episodic events. June 28 Conduct the S-R analysis between the diel range of DO and chlorophyll and percent saturation and chlorophyll to see if there is a relationship between these metrics and chlorophyll. June 28 Conduct the S-R analysis for the percent biovolume of individual taxa-producing species versus chlorophyll. June 28 Calculate a logistic regression for the cyanobacteria cell count versus chlorophyll S-R relationship, plot the cumulative proportion of cell count exceedances against chlorophyll, and compare the results of both calculations for Science Panel review. June 28 DWQ Confirm whether the "N/A" samples from the pH-chlorophyll S-R analysis are surface grab or composite samples. June 28 DECISIONS AND APPROVALS No formal decisions or approvals were made at this meeting. 2 SCIENCE PANEL DIRECTION ON STRESSOR-RESPONSES ANALYSES Throughout the meeting, the Science Panel provided direction to Tetra Tech on how to proceed with the stressor-response (S-R) analyses. The direction they provided is summarized below. • Science Panel members supported proceeding with the S-R analyses using the water quality data from the entire period of record. • Science Panel members indicated that more exploratory analysis is desirable to determine whether Goshen Bay should be considered a unique third region in the S-R analysis. At this time, Science Panel members supported proceeding with the S-R analyses by having Tetra Tech analyze Goshen Bay as a main part of the lake. The Science Panel can revisit that decision once more exploratory analyses are available. • Science Panel members supported assessing the following zooplankton metrics of interest against chlorophyll: percent Cladoceran, total biomass, and zooplankton production:biomass (P:B) ratio. One consideration in this analysis is that low Cladoceran levels (and high copepod levels) may be related to predatory pressure or the quality of the zooplankton food (i.e., phytoplankton taxa). The S-R analyses will help the Science Panel investigate these relationships further. • The Science Panel supported having Tetra Tech develop S-R visualizations using surface grab and composite sample data. The Science Panel will then use the S-R visualizations to continue to discuss whether to use surface grab versus composite samples and how to process the data by site type (i.e., whether to build an open water-to-nearshore translator). The S-R analysis may provide additional insight into the relationship between open water and nearshore data and if there is a difference between beach and marina sites, considering that marinas are enclosed. • For future discussion, the Science Panel supported proceeding with the S-R analyses using Tetra Tech’s recommendations based on preliminary S-R results, with additional direction to: o Conduct an S-R analysis for the diel range of DO vs. chlorophyll and percent saturation vs. chlorophyll. o Calculate the percent biovolume of individual taxa-producing species versus chlorophyll to explore whether there may be a more effective way to characterize the S-R relationship o Calculate a logistic regression for the cyanobacteria cell count versus chlorophyll S- R relationship and plot the cumulative proportion of cell count exceedances against chlorophyll o Include takeaways from the cyanobacteria relative biovolume versus chlorophyll S- R model in the S-R analysis report (i.e., that Provo Bay has a relatively lower representation of cyanobacteria than the main basin and that cyanobacteria are a general component of the phytoplankton community, irrespective of chlorophyll) • Science Panel members supported proceeding with the S-R analysis by having Tetra Tech compare the Utah Lake site-specific S-R models to the national models to see if they are consistent. 3 PROJECT ORIENTATION AND TECHNICAL SUPPORT DOCUMENT OVERVIEW Dr. Kateri Salk, Tetra Tech, provided an overview of available evidence and how it informs various aspects of the ULWQS project. Her overview is summarized below. Project Orientation and Technical Support Document (TSD) Overview Presentation Below is a summary of the project orientation and TSD overview presentation. Project Orientation Overview • Over the next two days, the ULWQS Science Panel will discuss the TSD, which is the document that will link management goals to response conditions and then to nutrients. The TSD ultimately will consist of the reference analysis, stressor-response (S-R) analysis, and literature evidence. • Once the TSD is complete, the Science Panel will develop protective numeric nutrient criteria (NNC) ranges and quantify uncertainty. They will report the protective nutrient ranges to the ULWQS Steering Committee. • While the Science Panel is developing the NNC recommendation, they will also be working on responses to the ULWQS Steering Committee charge questions. The Steering Committee first proposed the charge questions to learn more about the complex limnology of Utah Lake. In 2021, the Science Panel completed the first draft of the charge questions report, which provides responses to the questions. The Science Panel will revisit the charge questions report as new studies are completed. • Once the NNC recommendation is developed, the next step of the process is implementation planning. During this step, the Steering Committee and Science Panel will account for source partitioning, cost and feasibility, and timelines to develop a strategy to achieve the identified NNC. • As the Science Panel moves towards developing a recommended range of nutrients, members will consider multiple lines of evidence: the ULWQS studies, empirical S-R modeling, watershed mechanistic modeling, lake mechanistic modeling, and literature. The Science Panel will use each line of evidence in some capacity to develop the TSD reference analysis, S-R analysis, literature review, Steering Committee charge question responses, and implementation planning. • The pathway to criteria is outlined below: o The Science Panel will use the TSD to develop recommendations for protective nutrient ranges. o The ULWQS Steering Committee will consider the Science Panel's recommendations and make their own criteria recommendations. o The Steering Committee recommendations will be sent to the Utah Lake Commission for review and endorsement. o After the Utah Lake Commission's review, the recommendation will go into the regular process for adopting criteria. The Utah Water Quality Standards Work Group will review and decide whether to approve the criteria. o The Utah State Legislature and US Environmental Protection Agency (EPA) will review and decide whether to approve the criteria. o Throughout the process, the ULWQS Science Panel, Steering Committee, DWQ, and consultants will interact with groups, like the Water Quality Standards Work Group, to address needs. 4 TSD Overview • The purpose of the TSD is to provide the technical basis for developing NNC to protect designated uses (recreation, aquatic life, agriculture, and downstream). The technical consultants and Science Panel will conduct analyses to support multiple lines of evidence in the NNC framework. • The TSD will include three lines of evidence: o Reference-based evidence: Results from the Paleolimnological studies and the reference condition scenario under the Utah Lake Nutrient Model o S-R analysis: Statistical models and the Utah Lake Nutrient Model outputs o Scientific literature: Scientific studies of comparable/related lake ecosystems and other studies on Utah Lake • The analyses together will produce a range of nutrient levels deemed protective of uses across multiple lines of evidence. • Once a range is determined, the Science Panel will weigh the lines of evidence against each other to pick an appropriate threshold. They will do so by assessing the statistical distributions of endpoints and interpreting them in the context of their uncertainty. This step will involve assessing the lines of evidence based on their relevance, strength, and reliability. The Science Panel will use the established uncertainty matrix to assess the amount of evidence and the agreement of the evidence for the endpoints. S-R ANALYSIS – PERIOD OF ANALYSIS CONSIDERATION AND DISCUSSION One key consideration as part of the S-R analysis methodology is what period of record to use for the analysis. Dr. Kateri Salk, Tetra Tech, provided an overview of the different considerations associated with deciding what period to use for the analysis. Science Panel members discussed these considerations and provided direction on how to move forward with the S-R analysis. Her overview, the subsequent Science Panel discussion, and public comments are summarized below. Tetra Tech Overview of the Period of Analysis Consideration • There is a large amount of data available for Utah Lake. The Science Panel will need to decide how to process the data. One of the tradeoffs that the Science Panel will need to consider is whether to use all available data in the S-R analysis or more recent data (after 2015). Tradeoffs associated with this consideration include: o Using all available data results in a larger sample size and potentially larger gradient. o Using only recent data results in the application of consistent analytical methods and more intensive sampling (i.e., sampling has occurred more regularly within recent years). • Whether to use all available or more recent data is only relevant to constituents for which data has been collected over a long period. • The goal of the S-R analysis is to determine how the lake behaves, not to characterize recent conditions of the lake. • Analytical methods to measure total phosphorus (TP) changed throughout the period of record. The change in analytical methods has resulted in differences in minimum detection and minimum reporting limits over time. Other constituents have had more consistent methods over time, so considerations related to the change in analytical methods primarily apply to TP data. • The Science Panel Criteria Development Subgroup, which met to provide initial direction on the S-R analysis, suggested Tetra Tech create box plots to compare the data distribution pre-2015 and post-2015. Tetra Tech created box plots to compare the pre and post-2015 5 data for chlorophyll a, Secchi depth, TP, and total nitrogen (TN). The box plots indicate no substantial differences in the pre and post-2015 data. Furthermore, there were only three TP non-detects in either period, which removes concerns about the analytical methods change over time. It was difficult to compare pre and post-2015 TN data as TN was not extensively sampled before 2015. • Tetra Tech also created exploratory visualization of bivariate relationships between Secchi depth-TP, Secchi depth-TN, chlorophyll a-Secchi depth, chlorophyll a-TP, and chlorophyll a- TN. The takeaway from these exploratory visualizations is that the data pre-2015 generally overlapped with the post-2015 data. • The initial recommendation is to include data from the entire period of record instead of focusing on recent data. Tetra Tech will ultimately take direction from the Science Panel on whether to include the data from the entire period of record or from 2015 to the present. Science Panel Clarifying Questions Science Panel members asked clarifying questions about the assessments conducted on the analysis period. Their questions are indicated below in italics, with the corresponding responses in plain text. Is there a times series graph showing the data for the different water quality metrics collected at sites? No, but Tetra Tech could produce that time-series graph if it would be helpful to the Science Panel. Do the pre and post-2015 comparative graphs include data from all available sites in the analysis period? Yes, it does. Science Panel Discussion on the Period of Analysis Consideration • A time series graph that separated data collected from Provo Bay and data collected from the main lake over time may help increase understanding of how pre and post-2015 data differs across time and sites. Overall, Provo Bay and Goshen Bay are sampled less than the main lake due to access issues. • Some TN values appear to overlap less in the bivariate relationship graphs. Tetra Tech could run statistics to see if there is a significant difference, but it may not be significant simply because there are few TN samples before 2015. Public Comment on the Period of Analysis Consideration The amount of sediment coming into Utah Lake can be variable from year to year. Using a longer period of record will better capture the broad variability in conditions Utah Lake can experience annually. Science Panel Direction • Science Panel members supported using the water quality data from the entire period of record for the S-R analyses. • Tetra Tech will create a time series graph for different water quality variables across sites to explore the influence of episodic events. S-R ANALYSIS – SPATIAL AGGREGATION CONSIDERATION AND DISCUSSION One key consideration as part of the S-R analysis methodology is how to spatially aggregate data. Dr. Kateri Salk, Tetra Tech, provided an overview of the different considerations associated with deciding how to spatially aggregate data. Science Panel members discussed these considerations 6 and provided direction on how to move forward with the S-R analysis. Her overview, the subsequent Science Panel discussion, and public comments are summarized below. Tetra Tech Overview of the Spatial Aggregation Consideration • There are many sampling sites spread out across Utah Lake. • According to the State of Utah's Consolidated Assessment and Listing Methodology (CALM), the methodology by which Utah Lake is assessed for attainment, all uses must be met in all sampled locations. • The Science Panel Criteria Development Subgroup suggested analyzing the S-R relationship at the site scale to identify the most sensitive areas of the lake and align with the CALM methodology. • Another option is to aggregate up to a waterbody scale. If the Science Panel proceeded with this option, the aggregation would need to demonstrate that the lake-wide condition is protective of conditions at every site, which is the more complicated approach. • Utah Lake has two Assessment Units: the main basin and Provo Bay. One outstanding question is whether to consider Goshen Bay as a third distinct region. To define Goshen Bay as a third distinct region, the Science Panel would need to demonstrate that Goshen Bay has unique hydrology. • One challenge with characterizing Goshen Bay as a unique third region is that there is not as distinct a boundary between Goshen Bay and the main basin as between Provo Bay and the main basin. • If the Science Panel wants to assess whether Goshen Bay functions as a third unique region of the lake, one approach would be to see if there are differences in total dissolved solids (TDS) between Goshen Bay and the main lake. Another approach would be to run the S-R analyses with the three regions and see if the relationships differ. If the relationships do not differ, that would be evidence to combine the data and set a single threshold. If the relationships do differ, that would be evidence to analyze each region separately and set multiple thresholds. • Tetra Tech plotted TDS against latitude to measure if there was a shift in values between the main basin and Goshen Bay. One site has been extensively sampled in Goshen Bay, which could serve as a logical "break" to define the Goshen Bay boundary. When TDS samples are plotted by site (main basin, Goshen Bay, and Provo Bay), it shows similar hydrological properties between Goshen Bay and the main basin, while Provo Bay shows distinct hydrological properties. • It may be helpful to look at a time series graph of TDS between the main basin, Goshen Bay, and Provo Bay to see if there are identifiably unique properties over time. • There is limited data for Goshen Bay compared to Provo Bay. • The initial recommendation is to run the S-R analysis at the site scale and to consider Goshen Bay as part of the main basin. There are tradeoffs to consider in including Goshen Bay as a unique region in the analysis versus combining it with the main basin. Tetra Tech will ultimately take direction from the Science Panel on how to spatially aggregate data for the S-R analysis. Science Panel Clarifying Questions Science Panel members asked clarifying questions about the considerations associated with determining how to spatially aggregate data. Their questions are indicated below in italics, with the corresponding responses in plain text. 7 Why would the main basin differ from Goshen Bay? The water level fluctuates in Utah Lake. When the lake is near compromise, the main basin and Goshen Bay could have similar TDS values, but Goshen Bay may become saltier when the elevation is low. A time series could help verify whether this dynamic is occurring. What compromises TDS in the main lake? TDS is primarily composed of sodium chloride and calcium carbonate. Is the soil saltier around Goshen Bay? Freshwater flows are primarily coming into Utah Lake from the east. There are no direct tributaries into Goshen Bay. Any inflows are primarily coming from salty groundwater. The combination of evaporation and no freshwater inflows impact Goshen Bay. Science Panel Discussion on the Spatial Aggregation Consideration • It will be difficult to draw a line for Goshen Bay because there is insufficient data near the potential boundary between Goshen Bay and the main basin. There is one sampling site distinctly situated in Goshen Bay, but there is a big gap between that site and the next sampling site in the main basin. • One potential way to identify the boundary between Goshen Bay from the main basin is to track average depth as a function of latitude. DWQ has bathymetry maps that may show where there is a change in average depth between Goshen Bay and Utah Lake that could serve as a distinct boundary. • Criteria must be met in all parts of Utah Lake, including Goshen Bay. If Goshen Bay does not have distinct hydrological characteristics, then there might not be a rationale to break it out for the S-R analysis. Public Comment on the Spatial Aggregation Consideration The southern third of Goshen Bay can completely dry up at different times of the year. Science Panel Direction Science Panel members indicated that more exploratory analysis is desirable to determine whether Goshen Bay should be considered a unique third region in the S-R analysis. At this time, Science Panel members supported proceeding with the S-R analyses by having Tetra Tech analyze Goshen Bay as a main part of the lake. The Science Panel can revisit that decision once more exploratory analyses are available. S-R ANALYSIS – ZOOPLANKTON METRICS OF INTEREST DISCUSSION One key consideration as part of the S-R analysis methodology is how to incorporate zooplankton metrics into the S-R analysis to establish a threshold for zooplankton communities that support the aquatic life use. Dr. Kateri Salk, Tetra Tech, provided an overview of the different considerations associated with deciding how to incorporate zooplankton data. Science Panel members discussed these considerations and provided direction on how to move forward with the S-R analysis. Her overview, the subsequent Science Panel discussion, and public comments are summarized below. Tetra Tech Overview of the Zooplankton Metrics of Interest • The S-R analysis aims to quantify the level of primary production associated with the zooplankton community that supports aquatic life use. • Following the direction from the Science Panel Criteria Development Subgroup, Tetra Tech and DWQ met with Dr. Lester Yuan from the EPA to discuss the Zooplankton National Model. They also reviewed Dr. David Richards' Multi-Index of Biological Integrity (MIBI) report and met with Dr. Mike Brett to discuss his co-authored 2022 report (Zhang et al.). 8 Lastly, they met with June Sucker Recovery Program staff to discuss what zooplankton conditions support June sucker restoration efforts. They reviewed all these reports to discuss what zooplankton community conditions are supportive of the aquatic life designated use. • The Zooplankton National Model assesses the relationship between total zooplankton biomass and chlorophyll. The relationship has a steeper slope at low chlorophyll levels, and then, as trophic transfer becomes less efficient at higher chlorophyll levels, the slope levels off. One approach to selecting a zooplankton community metric that supports the aquatic life designated use is to select a threshold that prevents this decoupling. Utah Lake is unique due to its large size and shallow depth, so one option would be to run the Zooplankton National Model at multiple slope thresholds to generate a range of chlorophyll targets. • Dr. David Richards' MIBI report captures over 38 zooplankton metrics of interest. The report suggests the use of diversity and biomass as primary indicators. These metrics have an important role in Utah Lake. It is unclear how to translate those metrics and select a threshold representative of the zooplankton conditions supportive of the aquatic life use designation. • The Zhang et al. (2022) study characterizes the zooplankton production:biomass (P:B) ratio. The theoretical backing for calculating this ratio is that there is a physiological control on zooplankton productivity during cold periods. During warm periods, there is a resource control on zooplankton productivity. The underpinning of the report is that there is a food web energy transfer based on temperature and primary production. Applying this model to Utah Lake would require converting zooplankton counts to biomass via literature-derived relationships. The Utah Lake P:B ratios in Utah Lake could then be compared against other lakes in the study to inform thresholds. • The Landom et al. June Sucker Report suggests that larger-bodied Cladocera (Daphnia) are indicators of a zooplankton community supportive of a healthy fish community. The study also found that the zooplankton community correlated with carp biomass. One potential metric of interest to use is Cladoceran relative biomass. • Potential metrics/approaches to link zooplankton metrics to chlorophyll concentrations include the slope of zooplankton versus phytoplankton biomass in the EPA Zooplankton National Model, the metrics identified in Dr. Richards MIBI, the P:B ratio from Zhang et al. (2022), or Caldoceran relative abundance. These analyses could also be interesting because a strong relationship between the indicators and chlorophyll concentrations could indicate support for bottom-up controls linked to nutrients. In contrast, a weak relationship could indicate support for top-down controls or other confounding factors. • Tetra Tech requests input from the Science Panel on the metrics of interest, the possible approach to identify thresholds for the metrics of interest, and the method for weighing the value of the zooplankton analysis against other lines of evidence. Science Panel Clarifying Questions Science Panel members asked clarifying questions about the considerations associated with determining zooplankton metrics that can be used to establish a threshold supportive of the aquatic life use designation. Their questions are indicated below in italics, with the corresponding responses in plain text. The Zooplankton National Model appears to extrapolate far behind the high and low ranges of the sample data. Why does the plot extend so before the high and low values of the dataset? The visualization shown is an example of a specific region. There is data for other regions that extends beyond the displayed example of how to apply the Zooplankton National Model. 9 Is it appropriate to link the P:B ratio to chlorophyll concentrations since chlorophyll concentrations drive that ratio? Linking the P:B ratio back to chlorophyll concentrations likely would conclude that high cyanobacteria levels are undesirable. Calculating P:B ratios essentially quantifies the penalty associated with high cyanobacteria biomass on zooplankton production, as zooplankton tend not to favor cyanobacteria in their diet. What are the designated uses related to zooplankton? The primary use related to zooplankton is the aquatic life use designation. The Steering Committee was particularly interested in what constitutes a healthy diet for June sucker. The question for the Science Panel is whether chlorophyll can be managed to support a healthy June sucker population. Mike Mills and researchers in the June Sucker Recovery Program will be able to elaborate on what conditions and diet June sucker need. The Program is currently finalizing its 2023 report, which may be able to provide additional data. Science Panel Discussion on Zooplankton Metrics of Interest • The Zhang et al. (2022) study's conclusions are similar to the EPA's Zooplankton National Model in that the relationship between zooplankton and primary production decouples as chlorophyll levels get high. The researchers suggest that this decoupling occurs because the algae at these higher chlorophyll levels are lower-quality food. • The results of the Timpanogos Special Service District's (TSSD) Limnocorral and Mesocosm Studies indicate that the carp population is filtering out larger zooplankton populations. The Zhang et al. (2022) and Zooplankton National Model do not address the other biotic components (e.g., carp populations) impacting zooplankton populations and size distribution. • While the Zooplankton National Model is dependent on predation, the P-B ratio can be calculated independent of predation. The P:B ratio model is fairly simple and accounts for the quality of phytoplankton, type of phytoplankton, and temperature. The P:B ratio model calculates the percent growth per day for Daphnia and copepods. The data on temperature and breakdown of phytoplankton is available in Utah Lake to derive the P:B ratio for copepods. • Whether zooplankton consumes cyanobacteria depends on the size of the cells. Tiny copepods cannot eat larger cyanobacteria. • Studies indicate that Daphnia receives over half of its resources by consuming diatoms and cryptophytes. While Daphnia may eat cyanobacteria, they need access to more nutritional foods (i.e., diatoms and cryptophytes). • Tetra Tech could calculate the P:B ratio by translating total zooplankton and relative abundance data into biomass and using existing data on water temperature and phytoplankton community structures. • A high diversity of zooplankton may indicate that the zooplankton community is unhealthy. A healthy zooplankton community is dominated by Cladocera. A dominant Cladocera community is indicative of a healthy food web; a highly diverse zooplankton community may indicate that superior competitors, like Cladocera, have been pushed down. • The Utah Lake zooplankton community does not include a large Cladocera population, potentially due to fish predation. 10 Public Comment on Zooplankton Metrics of Interest For two decades, the June Sucker Recovery Program focused on increasing June sucker populations by reducing common carp populations. In response, the Program removed carp in large numbers. As the Program removed carp, the carp that remained in Utah Lake began to increase in size and reproduce. The June suckers now compete with stronger young carp. The carp will continue to eat whatever they can, including the large zooplankton. Science Panel Direction The zooplankton metrics of interest include percent Cladoceran, total biomass, and P:B ratio. Science Panel members supported assessing the following zooplankton metrics of interest against chlorophyll: percent Cladoceran, total biomass, and zooplankton production:biomass (P:B) ratio. One consideration in this analysis is that low Cladoceran levels (and high copepod levels) may be related to predatory pressure or the quality of the zooplankton food (i.e., phytoplankton taxa). The S-R analyses will help the Science Panel investigate these relationships further. S-R ANALYSIS – WATER QUALITY SAMPLING DATA DISCUSSION One key consideration as part of the S-R analysis methodology is how to process data between surface grab and composite samples and among different site types (open water, beaches, marinas). Dr. Kateri Salk, Tetra Tech, provided an overview of the considerations associated with deciding how to incorporate surface grab and composite sample data and data by site type into the S-R analysis. Science Panel members discussed these considerations and provided direction on how to move forward with the S-R analysis. Her overview, the subsequent Science Panel discussion, and public comments are summarized below. Tetra Tech Overview of Water Quality Sampling Data Challenges • There are several spatial components to the collected phytoplankton data. First, some samples were collected as surface "scum," while others were collected as a surface composite. The sampling routine for collecting surface composite samples was to collect samples "elbow deep" and run the sample through the photic zone. • One suggestion is to change the terminology from "surface scum" samples to "surface grab" samples. The reason for this change in terminology is that some surface grab samples were taken even though scum was not present on the surface. • The question for the Science Panel is whether to interpret surface grab samples differently or the same as surface composite samples. • Tetra Tech created box plots to compare cyanobacteria cell counts for surface grab and composite samples across different sampling locations (beach, marina, open water – main basin, and open water – Provo Bay). o The cyanobacteria cell counts in the surface grab samples were generally higher than in the composite samples. o The distributions of the cyanobacteria cell counts were similar across site types for composite samples. The open-water sites generally had lower cyanobacteria cell counts than the beach/marina for surface grab samples but had smaller sample sizes. • The patterns in the cyanobacteria cell count data are similar to those in the cyanobacteria biovolume data. • The microcystin concentrations were generally higher in the surface grab samples than in the composite samples. The marina samples tended to have higher toxin concentrations than other site types. 11 • Tetra Tech paired the surface grab samples to composite samples on a bivariate plot for cyanobacteria cell counts. When paired, the surface grab samples always exceeded the composite sample. The samples were almost exclusively restricted to beach and marina samples. • Tetra Tech paired the surface grab samples to composite samples on a bivariate plot for cyanobacteria biovolume. When paired, the surface grab samples always exceeded the composite samples in beach and marina samples. In the main basin open water sites, the relationship straddles the 1:1 line. • Tetra Tech paired the surface grab samples to composite samples on a bivariate plot for microcystin concentrations. When paired, the surface grab samples always exceeded the composite sample. The paired samples were almost exclusively restricted to beach and marina samples. • Another spatial component complicating the phytoplankton data is that sampling locations are found in the open water, along the beach, and in marinas. One question for the Science Panel to consider is how to interpret the results of the phytoplankton samples across these spatial differences considering that uses must be attained in all areas of the lake. • One approach for assessing water quality conditions across the lake is to use a "translator" approach for open water versus nearshore samples. The translator approach would potentially identify the open water conditions needed to support designated uses in beaches/marinas. If the translator approach provided evidence of what open water conditions are needed to support designated uses in beaches/marinas, the next step would be to run the S-R models for open water sites to translate to nutrient concentrations. • Tetra Tech compared the open water data to beach data and marina data on a bivariate plot for cyanobacteria cell counts, cyanobacteria biovolume, and microcystin concentrations to see if there was a relationship between open water and nearshore samples. The overall analysis indicated no systematic bias between site types when samples were matched in time for cyanobacteria biovolume and cell counts. However, beach and marina samples tend to be higher for microcystin concentrations than open water sites. Tetra Tech recommends building the translator. If the translator shows a systematic bias, then the recommendation is to use it to identify open-water targets to protect nearshore sites. If there is no systematic bias, then the recommendation is to treat sites as consistent with one another. • Tetra Tech requests Science Panel input on whether to treat surface grab and composite samples similarly or differently and whether they support the recommendation to develop an open water to nearshore translator. Science Panel Clarifying Questions Science Panel members asked clarifying questions about the considerations associated with processing the sampling data. Their questions are indicated below in italics, with the corresponding responses in plain text. When were composite samples collected compared to surface grab samples? The composite samples are collected once a month. The surface grab samples are mostly collected as part of the Harmful Algal Bloom (HAB) Advisory Program. A composite sample is collected whenever a surface grab sample is collected as part of the HAB Advisory Program. Are the beaches and marinas sampled routinely? One of the challenges is that marinas and beaches are monitored differently than open water sites. The question for the Science Panel is whether they would assess marinas and beaches differently 12 for use attainment. Setting criteria for open water conditions that protect marinas and beaches will affect how use attainment is assessed in the lake. Marinas are often enclosed and separate from the rest of the lake. Can improving conditions in the open water result in the improvement of marina conditions? The Utah Lake Commission is conducting treatments on marinas separately. The open water and marinas are currently in less-than-ideal conditions. The Utah Lake Commission can continue to treat marinas separately to help improve water quality. Science Panel Discussion on Water Quality Sampling Data Challenges • Surface grab samples tell a researcher very little about the big-picture condition of the lake. Taxa are vertically migrating and mixing in the lake. Vertical migrations and diel mixing confound the results from the surface grab samples. • The surface grab and composite sample data measure different things. The surface grab samples measure how bad a bloom can be and are relevant to public health. The composite samples better capture the physical properties of the water column. If the S-R analysis aims to assess the relationship between chlorophyll and nutrients, the composite samples are a better measure of that relationship. • The surface grab samples can be useful if total phosphorus is measured in those samples, too. The surface grab samples collected by DWQ primarily do not have nutrients associated with them. DWQ has only recently begun including measuring nutrients in the surface grab samples. It is possible to link the surface grab samples to chlorophyll levels, but fewer opportunities exist to link them to nutrient levels. • The surface grab samples are relevant to public health considerations. Surface grab samples near the shoreline can estimate the potential risk of exposure for any recreators. Recreators may choose to avoid surface scum associated with the surface grab samples. At the same time, lysed cells will release microcystin into the water, so it can be difficult to estimate the public health risk by the absence or presence of surface scum. • The microcystin concentrations in the surface grab samples are an order of magnitude higher than the concentrations in the composite samples. • A potential next step is to conduct the S-R analysis with all the data included, so Science Panel members can better visualize how these different factors (e.g., site type, surface grab vs. composite samples) affect the S-R analysis. It will be easier to decide how to proceed if there are visualizations. Public Comment on the Water Quality Sampling Data Challenges • There is little interaction between the open lake and the marinas. Additionally, the wind can blow cyanobacteria mats to beaches; these mats then decay, increasing exposure and impacting recreation. • One question to consider is whether marinas could be considered a pollution source. Science Panel Direction The Science Panel supported having Tetra Tech develop S-R visualizations using surface grab and composite sample data. The Science Panel will then use the S-R visualizations to continue to discuss whether to use surface grab versus composite samples and how to process the data by site type (i.e., whether to build an open water-to-nearshore translator). The S-R analysis may provide additional insight into the relationship between open water and nearshore data and if there is a difference between beach and marina sites, considering that marinas are enclosed. 13 PRELIMINARY S-R ANALYSIS OVERVIEW Dr. Kateri Salk, Tetra Tech, presented an overview of the S-R analysis. Her presentation is summarized below. • The S-R analysis intends to establish statistical relationships between stressors and responses. The responses are linked to management goals and designated use support. The S-R analysis answers, "What stressor level would relate to a response level that supports the use?" • There will be variability in the results, so the analysis will also incorporate an assessment of statistical uncertainty. Additionally, confounding factors can be incorporated as covariates or contribute to statistical uncertainty. • The S-R analyses pertain to water column conditions. The relationships are chosen based on causal linkages. The relationships are independent of nutrient sources. The S-R analysis aims to illustrate what is occurring in the water column. The ULWQS Steering Committee, with support from the Science Panel, can determine later how to achieve those conditions as part of the implementation planning process. • The general approach to the S-R analysis is to filter the dataset for surface samples collected from April to September. (The period may be extended to October.) Today's preliminary results include surface grab sample data and do not include time period restrictions. Each site-date observation is considered a single sample in the analysis. • The initial approach to the S-R analysis was to test whether a linear regression appropriately fits the dataset. If the analysis fits the assumption of linear regression, Tetra Tech recommends proceeding with that approach. If the analysis does not fit the assumptions of linear regression, Tetra Tech recommends a different approach (e.g., quantile). • Tetra Tech tested the statistical significance for each stressor variable. They also delineated the data between the main basin and Provo Bay to statistically test whether there was a difference between the main basin and Provo Bay. • In the future, the Science Panel will discuss and provide guidance on what confident/credible intervals and prediction intervals/quantiles represent quantification of uncertainty and protectiveness. For example, the Science Panel will consider whether to protect against exceedances in 95% of samples, 90% of samples, etc. PRELIMINARY S-R ANALYSIS RESULTS: DISSOLVED OXYGEN-CHLOROPHYLL ANALYSIS Dr. Kateri Salk, Tetra Tech, presented an overview of the preliminary S-R analysis results between dissolved oxygen (DO) and chlorophyll. Science Panel members discussed the preliminary results and provided direction on how to move forward. Her overview, the subsequent Science Panel discussion, and public comments are summarized below. DO-Chlorophyll S-R Analysis Overview • Tetra Tech conducted S-R analyses that linked the 30-day mean DO, the 7-day mean, and the 1-day minimum DO values to chlorophyll. • The DO values are derived from continuous buoy measurements from four Utah Lake locations. They are then paired with chlorophyll grab samples, observed on the same day. The S-R analysis only uses DO values when there is a corresponding chlorophyll value. • 30-Day Mean DO vs. Chlorophyll: The S-R model between the 30-day mean DO and chlorophyll indicated that chlorophyll was not a significant predictor of the 30-day mean DO. Additionally, Utah Lake rarely observed 30-day mean DO exceedances, except in Provo Bay, where two exceedances were observed. The 30-day mean DO versus chlorophyll S-R 14 model is not a promising indicator for generating a chlorophyll target, so Tetra Tech recommends not using it. • 7-Day Mean DO vs. Chlorophyll: There are two 7-day mean DO thresholds for determining exceedances: a threshold for normal biotic conditions and a threshold if early life stages are present. The S-R model between the 7-day mean DO and chlorophyll indicated that chlorophyll is not a significant predictor of 7-day mean DO in the main basin. The model did show a negative relationship between chlorophyll and 7-day mean DO, but the sample size was too small to demonstrate significance. Utah Lake rarely observed exceedances, except in Provo Bay. The 7-day mean DO versus chlorophyll S-R model is not a promising indicator for generating chlorophyll targets, so Tetra Tech recommends not using it. • 1-Day Minimum DO vs. Chlorophyll: There are two 1-day minimum DO thresholds: a threshold for normal biotic conditions and a threshold if early life stages are present. The S- R model between 1-day minimum DO values and chlorophyll indicated a significant negative relationship in the main basin; still, no exceedances were observed in the main basin during the period of record. There were 1-day minimum DO value exceedances observed in Provo Bay, particularly if the higher criterion for early life stages is in place. The S-R analysis for Provo Bay has a significantly lower intercept than the analysis for the main basin. Still, the slope between the Provo Bay and main basin data is not significantly different. Tetra Tech recommends generating chlorophyll targets using the linear regression between 1-day minimum DO values and chlorophyll and a separate intercept term for Provo Bay. Science Panel Clarifying Questions on the DO-Chlorophyll S-R Analysis Science Panel members asked clarifying questions about the DO-Chlorophyll S-R analysis. Their questions are indicated below in italics, with the corresponding responses in plain text. Did Tetra Tech consider conducting the S-R analysis using the diel range for DO against chlorophyll? Tetra Tech only conducted the S-R analyses for the 30-day mean, 7-day mean, and daily minimum. They could look at the diel range if that interests the Science Panel. Does the S-R analysis include all the data through 2022? Yes, the S-R analysis includes all the data through 2022. There are fewer points on the graph than expected because chlorophyll grab samples are taken monthly, and the analysis only uses points where both chlorophyll and DO values were observed. A relationship between chlorophyll and 30-day mean DO would not be expected because a 30-day mean value would account for diurnal swings in DO. If the purpose of the S-R model is to generate a chlorophyll target, would it not be more advantageous to use metrics like the diel range of DO or percent saturation? There is precedent for using the diel range of DO values or percent saturation, particularly in streams. Ultimately, there is more ecological backing for using a 1-day minimum rather than a 30- day mean, which would be obscured by diel fluctuations in DO. Science Panel Discussion on DO-Chlorophyll S-R Analysis • The national EPA model predicts the influence of chlorophyll on DO. The national EPA model is irrelevant to Utah Lake because the model is based on deeper, stratified lakes. • Tetra Tech can conduct an S-R analysis between the diel range of DO and percent saturation and chlorophyll to see if there is a relationship between these metrics and chlorophyll. These metrics may serve as a signal for elevated nutrient concentrations, particularly for Provo Bay. 15 Public Clarifying Questions on the DO-Chlorophyll S-R Analysis Members of the public asked clarifying questions about the DO-Chlorophyll S-R analysis. Their questions are indicated below in italics, with the corresponding responses in plain text. At what time of year do the DO exceedances in Provo Bay occur? The DWQ's buoys collect data during the summer, so the exceedances occur sometime in the summer. The exact month is uncertain, but that is information that Tetra Tech can identify. Are early life stages present in the late summer (i.e., August)? There is a separate DO criterion for when early life stages are present. There was a past effort to identify when early life stages are present in Utah Lake. DWQ has that information and will need to revisit it to provide a more definitive answer. Public Comment on the DO-Chlorophyll S-R Analysis Low DO levels generally do not impact carp populations because they can respire at the surface. June suckers are more threatened by DO levels, particularly in their early life stages. Science Panel Direction on DO-Chlorophyll S-R Analysis The Science Panel supported proceeding with the S-R analyses using Tetra Tech's recommendations for the 30-day mean DO vs. chlorophyll, 7-day mean DO vs. chlorophyll, and 1- day minimum DO vs. chlorophyll S-R relationship, with additional direction to conduct an S-R analysis for the diel range of DO vs. chlorophyll and percent saturation vs. chlorophyll. PRELIMINARY S-R ANALYSIS RESULTS: DAILY MAXIMUM PH-CHLOROPHYLL ANALYSIS Dr. Kateri Salk, Tetra Tech, presented an overview of the preliminary S-R analysis results between daily maximum pH and chlorophyll. Science Panel members discussed the preliminary results and provided direction on how to move forward. Her overview, the subsequent Science Panel discussion, and public comments are summarized below. Daily Maximum pH-Chlorophyll S-R Analysis Overview • There is a daily minimum pH threshold that Utah Lake rarely hits due to it being a highly alkaline lake with high pH. There is a daily maximum pH threshold that functions as a "not to exceed" threshold. • Daily Max pH vs. Chlorophyll: The S-R model indicates a positive relationship between chlorophyll and pH in the main basin. There is a negative relationship observed in Provo Bay, but that is based on a sample size of six. The R2 value is 0.52. Tetra Tech recommends generating chlorophyll targets using this linear regression, either for the main basin only or by combining the data with Provo Bay, given the small sample size. Science Panel Discussion on Daily Maximum pH-Chlorophyll S-R Analysis • In Provo Bay, DWQ collects pH simultaneously as chlorophyll when they conduct grab samples. Using data from the grab samples in Provo Bay could help extend the period of record and increase the number of observations in the S-R analysis for the daily maximum pH and DO percent saturation. • Some of the paired pH-chlorophyll data points are surface grab samples paired with buoy data. The surface grab and composite samples are included in the S-R model. • The data points labeled in the S-R plot as "N/A" are from routine samples. DWQ can confirm whether the "N/A" samples are composite or surface grab. 16 Science Panel Direction The Science Panel supported proceeding with the S-R analyses using Tetra Tech’s recommendation for the daily max vs. chlorophyll S-R analysis. PRELIMINARY S-R ANALYSIS RESULTS: MICROCYSTIN-RESPONSE VARIABLE ANALYSIS Dr. Kateri Salk, Tetra Tech, presented an overview of the preliminary S-R analysis results between microcystin and different response variables, including cyanobacteria biovolume, cyanobacteria cell count, and chlorophyll. Science Panel members discussed the preliminary results and provided direction on how to move forward. Her overview, the subsequent Science Panel discussion, and public comments are summarized below. Microcystin-Response Variable S-R Analysis Overview • Microcystin vs. Cyanobacteria Biovolume: The S-R model for microcystin versus cyanobacteria biovolume indicates a significantly positive relationship. Provo Bay did not have a significantly different slope or intercept compared to the main basin's slope or intercept. The R2 value was 0.26. Tetra Tech could use a statistical approach other than linear regression to measure significance. There is a large spread among data, particularly as cyanobacteria biovolume reaches a higher level; the data suggests that exceedances are tied to high biovolumes, but high biovolumes do not always result in exceedances in microcystin. Tetra Tech recommends generating cyanobacteria biovolume targets using this linear regression, combining the main basin and Provo Bay data. • Microcystin vs. Cyanobacteria Cell Count: The S-R model for microcystin versus cyanobacteria cell counts indicates a significantly positive relationship. Provo Bay did not have a significantly different slope or intercept compared to the main basin's slope or intercept. Like the microcystin-cyanobacteria biovolume S-R analysis, the microcystin exceedances occur at high cyanobacteria cell count levels, but high cyanobacteria cell counts levels do not always result in exceedances. Tetra Tech recommends generating cyanobacteria cell targets using this linear regression, combining the main basin and Provo Bay data, and comparing them to the existing cell count thresholds. • Microcystin vs. Chlorophyll: The S-R model for microcystin versus chlorophyll indicates no significant relationship. The linkage between toxin production and chlorophyll is less direct than the cyanobacteria metrics. The linkage between may toxin production and chlorophyll may be less direct because of the limited sample size. It may be helpful to compare the output of the microcystin-chlorophyll S-R model to the EPA national model for microcystin to see if there is a consistency between the two models. The microcystin- chlorophyll S-R model is not a promising indicator for generating a chlorophyll target, so Tetra Tech recommends not using it. Science Panel Clarifying Questions on the Microcystin-Response Variable S-R Analysis Science Panel members asked clarifying questions about the microcystin-response S-R analysis. Their questions are indicated below in italics, with the corresponding responses in plain text. If microcystin is used as an indicator, would the criterion threshold be eight mg/mL? Utah has not adopted specific criteria for microcystin. The eight mg/mL threshold is the EPA's criteria. The State of Utah is currently assessing microcystin criteria. The ULWQS Steering Committee included eight mg/mL as one of the management goals. Science Panel Discussion on Microcystin-Response Variable S-R Analysis • The small R2 value for the microcystin-cyanobacteria biovolume S-R analysis may be because several cyanobacteria taxa do not produce microcystin, including picoplankton. 17 • In the microcystin-cyanobacteria cell count S-R model, the data points that exceed the microcystin threshold diverge most from the line of best fit. There is noise in the upper range of the microcystin-cyanobacteria cell count data; the Science Panel will need to consider this noise when discussing risk management and the likelihood of exceedances. • It is important to protect beaches as that is where most people are recreating, so it would be reasonable to use samples with high microcystin values from scum near the beach. Many of the samples with high microcystin values come from open-water locations. The S-R microcystin-cyanobacteria cell count plots do not distinguish open water versus nearshore samples, but a plot like that may be helpful for investigative purposes. Public Comment on the Microcystin-Response Variable S-R Analysis It may be worth considering running a quantile regression or "broken stick" regression to account for the spread of the data points, particularly at higher cyanobacteria biovolumes. There can be subjectivity in a "broken stick" regression. The original EPA guidance is to use a linear approach when it fits the assumption. One option to reduce the noise at the higher end is to remove the surface grab samples. Science Panel Direction The Science Panel supported proceeding with the S-R analyses using Tetra Tech's recommendations for the microcystin vs. cyanobacteria biovolume, microcystin vs. cyanobacteria cell count, and microcystin vs. chlorophyll S-R analyses. PRELIMINARY S-R ANALYSIS RESULTS: CYANOBACTERIA METRICS-CHLOROPHYLL ANALYSIS Dr. Kateri Salk, Tetra Tech, presented an overview of the preliminary S-R analysis results between several cyanobacteria stressor metrics (e.g., cyanobacteria biovolume, cyanobacteria cell count, cyanobacteria relative biovolume) and chlorophyll. Science Panel members discussed the preliminary results and provided direction on how to move forward. Her overview, the subsequent Science Panel discussion, and public comments are summarized below. Cyanobacteria Stressor Variable-Chlorophyll S-R Analysis Overview • Cyanobacteria Biovolume vs. Chlorophyll: The S-R model for cyanobacteria biovolume versus chlorophyll indicates a significant positive relationship. Provo Bay did not have a significantly different slope or intercept compared to the main basin. Tetra Tech recommends generating chlorophyll targets using the linear regression and biovolume targets identified in the microcystin-cyanobacteria biovolume S-R model. They also recommend combining the main basin and Provo Bay. • Cyanobacteria Cell Count vs. Chlorophyll: The S-R model for cyanobacteria cell count versus chlorophyll indicates a significantly positive relationship. The Provo Bay data had a significantly different intercept but a similar slope to the main basin data. Overall, Provo Bay had a lower cell count per chlorophyll than the main basin. The Utah recreational advisory cell count threshold is 100,000 cells/mL. Tetra Tech recommends generating chlorophyll targets using this linear regression, using a different intercept for Provo Bay. • Cyanobacteria Relative Biovolume vs. Chlorophyll: The S-R model for cyanobacteria relative biovolume versus chlorophyll indicates a very noisy relationship. The Y-axis values range from zero to one because the cyanobacteria relative biovolume is expressed as a proportion of the total biovolume. Because the Y-axis values range from zero to one, linear regression is not the best statistical approach; a logistic regression may be more appropriate. However, a different statistical approach would not likely illuminate a pattern. The S-R model is not a promising indicator for generating a chlorophyll target, so Tetra Tech recommends not using it. 18 Science Panel Clarifying Questions on the Cyanobacteria Stressor Variable-Chlorophyll S-R Analysis Science Panel members asked clarifying questions about the cyanobacteria stressor variable- chlorophyll S-R analysis. Their questions are indicated below in italics, with the corresponding responses in plain text. The State of Utah uses cell counts to inform its HAB Advisory Program. They do not use cyanobacteria biovolume. How relevant is cyanobacteria biovolume as an indicator if there is no relationship to an existing metric or indicator? The S-R analysis will help set a cyanobacteria biovolume target based on the relationship between cyanobacteria biovolume and microcystin. The next step of the S-R analysis is to identify the chlorophyll target based on the relationship between cyanobacteria biovolume and chlorophyll. The S-R analysis assesses the relationship between cyanobacteria biovolume and chlorophyll. Given that other algae besides cyanobacteria can contribute to chlorophyll measurements, would it be more appropriate to assess phytoplankton biovolume versus chlorophyll? • The R2 value for the cyanobacteria biovolume versus chlorophyll S-R model is 0.25, indicating a statistical relationship between cyanobacteria biovolume and chlorophyll. • Tetra Tech conducted additional analyses examining the relationship between more specific cyanobacteria taxa (e.g., dolichospermum) and chlorophyll. The relationship was stronger as the analysis focused on more specific taxa. Although assigning thresholds by specific taxa may not be advantageous, these analyses may help respond to charge questions. Three taxa are known toxin-producers: dolichospermum, microcystis, and aphanizomenon. It would be helpful for the Science Panel to see the S-R analyses between the specific toxin-producing taxa and chlorophyll. If thresholds were set for specific toxin producers, would monitors have to measure toxin-producing species to assess attainment? Would it be possible to set a chlorophyll threshold above which monitors would assess samples for the presence and abundance of specific toxin producers? • There are no other examples of thresholds set by the presence of toxin-producing species. • One potential approach could be to conduct multiple S-R analyses and develop a range representing different phytoplankton categories. • If the Science Panel wants to process the data by individual taxa, it would be helpful to have a specific plan on what taxa to explore. Is the 100,000 cyanobacteria cells/mL a standard in Utah? The 100,000 cyanobacteria cells/mL is not a standard in Utah. That threshold is used for recreational advisories. Science Panel Discussion on the Cyanobacteria Stressor Variable-Chlorophyll S-R Analyses • One way to potentially pursue an S-R analysis to assess individual taxa is to plot the percent biovolume of toxin-producing taxa against chlorophyll. • One potential approach to assess the cyanobacteria cell count versus chlorophyll S-R relationship is to conduct a logistic regression rather than a linear regression. A logistic regression defines each sample as either exceeding (value of one) or not exceeding (value of zero). The statistical relationship in the logistic regression displays the likelihood of exceedance across the data. Another approach would be to plot the cumulative proportion of cell count exceedances against chlorophyll to visualize a continuum for how likely a sample is to pass the cell count threshold based on a given chlorophyll value. 19 • The S-R analysis for cyanobacteria relative biovolume versus chlorophyll is interesting because it shows that cyanobacteria are prevalent, even at lower chlorophyll levels. It also shows that Provo Bay has a relatively lower representation of cyanobacteria than the main basin. Green algae can sometimes dominate in really productive systems. This analysis shows that cyanobacteria are a general component of the phytoplankton community, irrespective of chlorophyll. The nosiness of the relationship is just as informative as finding a pattern in the data for characterizing the lake; the nosiness of the relationship is not useful for the S-R analysis. Public Comment on the Cyanobacteria Stressor Variable-Chlorophyll S-R Analysis • The appropriate statistical method would be fractional regression for the S-R models that plot the cumulative proportion of cell count exceedances against chlorophyll. • The appropriate statistical method for the cyanobacteria relative biovolume versus chlorophyll S-R model may also be fractional regression. Science Panel Direction on the Cyanobacteria Stressor Variable-Chlorophyll S-R Analyses • The Science Panel supported proceeding with the S-R analyses using Tetra Tech's recommendations for the cyanobacteria biovolume vs. chlorophyll, cyanobacteria cell count vs. chlorophyll, and cyanobacteria relative biovolume vs. chlorophyll S-R analyses, with the additional direction to: o Calculate the percent biovolume of individual taxa-producing species versus chlorophyll to explore whether there may be a more effective way to characterize the S-R relationship o Calculate a logistic regression for the cyanobacteria cell count versus chlorophyll S- R relationship and plot the cumulative proportion of cell count exceedances against chlorophyll o Include takeaways from the cyanobacteria relative biovolume versus chlorophyll S- R model in the S-R analysis report (i.e., that Provo Bay has a relatively lower representation of cyanobacteria than the main basin and that cyanobacteria are a general component of the phytoplankton community, irrespective of chlorophyll) PRELIMINARY S-R RESULTS: SECCHI DEPTH-CHLOROPHYLL, TN, TP ANALYSIS Dr. Kateri Salk, Tetra Tech, presented an overview of the preliminary S-R analysis results between Secchi depth and chlorophyll, TN, and TP. Science Panel members discussed the preliminary results and provided direction on how to move forward. Her overview, the subsequent Science Panel discussion, and public comments are summarized below. Secchi Depth-Chlorophyll, TN, TP S-R Analysis Overview • The linear regression shows significantly negative relationships between Secchi depth and chlorophyll, TN, and TP, but the relationships are noisy (R2=0.05 to 0.13). • The S-R relationship between Secchi depth and chlorophyll, TN, and TP is likely confounded by suspended sediments. • One question for the Science Panel is: Is there a Secchi depth threshold that would support aquatic life and recreation? Science Panel Discussion on the Secchi Depth-Chlorophyll, TN, TP S-R Analyses • Utah Lake has a high number of suspended sediments in the water column. According to a previous light extinction analysis that the Science Panel guided, Utah Lake is light-limited despite being shallow. 20 • The light limitation impacts phytoplankton growth; however, given the presence of the suspended sediments, many confounding variables make the relationship between Secchi depth and chlorophyll, total nitrogen, and total phosphorus noisy. • It may be helpful to plot Secchi depth against fixed suspended solids to see if there is a relationship between the two variables. The expectation is that inert suspended sediments compose the majority of the turbidity in Utah Lake. • There is a strong seasonal component to water clarity in Utah Lake. The lake is relatively clear in the spring and less so in the summer and fall. The seasonal variability in water clarity may be contributing to the noisiness of the relationship. Public Comment on the Secchi Depth-Chlorophyll, TN, TP S-R Analyses Increasing the clarity of Utah Lake will increase the temperature of the lake, which in turn will affect the fish. It is important to consider water temperature impacts on the beneficial uses. Provo Bay is different than the main basin because it is so shallow; there have not been fish die-offs in Provo Bay, while there have been fish die-offs in the main basin. Science Panel Direction on the Secchi Depth-Chlorophyll, TN, TP S-R Analyses The Science Panel supported proceeding with the S-R analyses using Tetra Tech’s recommendation to not use the Secchi depth versus chlorophyll, TN, TP S-R analyses to identify chlorophyll and nutrient targets. PRELIMINARY S-R RESULTS: STRESSOR VARIABLE-NUTRIENT ANALYSIS Dr. Kateri Salk, Tetra Tech, presented an overview of the preliminary S-R analysis results between stressor variables (cyanobacteria biovolume, cyanobacteria cell counts, and chlorophyll) and nutrient response variables (TN and TP). Science Panel members discussed the preliminary results and provided direction on how to move forward. Her overview, the subsequent Science Panel discussion, and public comments are summarized below. Stressor Variable-Nutrient S-R Analyses Overview • Cyanobacteria Biovolume vs. TN: The S-R model for cyanobacteria biovolume versus TN does not indicate a relationship between the two variables. Tetra Tech recommends not using this S-R model to generate a TN target. • Cyanobacteria Biovolume vs. TP: The S-R model for cyanobacteria biovolume versus TP indicates a significant positive relationship. Provo Bay did not have a significantly different slope or intercept compared to the main basin. There is variability within the S-R model; the N:P ratio may help explain the high variability in cyanobacteria biovolume at certain total phosphorus concentrations. Given the wedge-shaped relationship between cyanobacteria and TP, quantile regression may be a more appropriate statistical method than linear regression. Tetra Tech recommends using the cyanobacteria biovolume vs. TP S-R relationship to derive TP targets, combining the main basin and Provo Bay data, and potentially switching to quantile regression. • Cyanobacteria Cell Count vs. TN: The S-R model for cyanobacteria cell count versus TN indicates a significant positive relationship, but the analysis is noisy. There is no significant difference in slope or intercept between Provo Bay and the main basin. Linear regression may not be the appropriate method for statistically analyzing the relationship; quantile regression may be more appropriate given the wedge-shaped relationship. Some of the noise in the S-R model may be due to changes in nutrient limitations between nitrogen and phosphorus. Tetra Tech recommends using the cyanobacteria cell count vs. TN S-R 21 relationship to derive TN targets, acknowledging uncertainty, combining the main basin and Provo Bay data, and potentially switching to quantile regression. • Cyanobacteria Cell Count vs. TP: The S-R model for cyanobacteria cell count versus TP indicates a significant positive relationship, but the analysis is noisy. There is no significant difference in slope or intercept between Provo Bay and the main basin. Quantile regression may be a more appropriate statistical approach given the wedge-shaped relationship of the dataset. Tetra Tech color-coded the data points to display the N:P ratios; the results show that higher N:P ratios occur at lower TP concentrations, and lower N:P ratios occur at higher TP concentrations. Tetra Tech recommends using the cyanobacteria cell count vs. TP S-R relationship to derive TP targets, acknowledging uncertainty, using a different intercept for Provo Bay, and potentially switching to quantile regression. • Chlorophyll vs. TN: The S-R model for chlorophyll versus TN indicates a significant positive relationship. Provo Bay had a significantly different slope and intercept than the main basin. The S-R model shows a wedge-shaped relationship, so quantile regression may be a more appropriate statistical method than linear regression. The N:P ratio may help explain variability. Tetra Tech recommends using the chlorophyll vs. TN S-R relationship to derive TN targets, using a different slope and intercept for Provo Bay, and potentially switching to quantile regression. • Chlorophyll vs. TP: The S-R model for chlorophyll versus TP indicates a significant positive relationship. Provo Bay did not have a significantly different slope and intercept than the main basin. The S-R model shows a wedge-shaped relationship, so quantile regression may be a more appropriate statistical method than linear regression. The N:P ratio may help explain variability. Tetra Tech recommends using the chlorophyll vs. TP S-R relationship to derive TP targets, combining main basin and Provo Bay data, and potentially switching to quantile regression. • As the Science Panel thinks about covariates (nitrogen and phosphorus) to explain S-R relationships, it is important to consider that nitrogen and phosphorus values are not associated with every sample. It is also important to consider if different conditions would result in different thresholds. Science Panel Clarifying Questions on the Stressor Variable-Nutrient S-R Analyses Science Panel members asked clarifying questions about the stressor variables (cyanobacteria biovolume, cyanobacteria cell counts, and chlorophyll) and nutrient response variables (TN and TP) S-R analyses. Their questions are indicated below in italics, with the corresponding responses in plain text. Was any cyanobacteria cell count-TN S-R model data from the ULWQS Bioassay Study? No, the data points displayed are all from direct monitoring. The Bioassay Study may provide insights into the conditions under which nitrogen and phosphorus are limiting but less so on light limitations. In the chlorophyll versus TP S-R model, the current statistical method uses a log-linear fit. Would it make more sense to use an asymptotic fit instead? Using a quantile regression over a linear regression would address some of these concerns. If there is a non-linear response, the Science Panel should consider using a statistical method that assumes a non-linear relationship. 22 Science Panel Discussion on the Stressor Variable-Nutrient S-R Analyses • The S-R analysis presumes that TP is causing blooms, but it is also plausible that blooms greatly increase the lake's phosphorus. In Upper Klamath Lake, blooms bring phosphorus into the water column. • One of the dynamics observed in Upper Klamath Lake is that the conditions are anoxic at the bottom of the lake, which results in the release of iron-bound phosphorus. When the wind dies down, the cyanobacteria migrate to the bottom of the lake and take up the phosphorous. • One approach to assess whether covariates influence the noise is to evaluate the residuals on the linear regression or develop a high-yield approach. The high-yield approach may help identify what is driving chlorophyll among the covariates. Science Panel Direction on the Stressor Variable-Nutrient S-R Analyses The Science Panel supported proceeding with the S-R analyses using Tetra Tech's recommendations for the cyanobacteria biovolume vs. TN, cyanobacteria biovolume vs. TP, cyanobacteria cell count vs. TN, cyanobacteria cell count vs. TP, chlorophyll vs. TN, and chlorophyll vs. TP S-R analyses. OUTSTANDING S-R ANALYSES There are several outstanding S-R analyses yet to be conducted. Dr. Kateri Salk, Tetra Tech, and Scott Daly, DWQ, presented the remaining S-R analyses. Science Panel members discussed the outstanding S-R analyses and provided direction on how to move forward. Their overview, the subsequent Science Panel discussion, and public comments are summarized below. Outstanding S-R Analyses Overview • There are several national models to which the Science Panel can compare the Utah Lake S- R analyses. It may be useful to compare the site-specific S-R models to the national microcystin-chlorophyll, zooplankton-chlorophyll, TN-chlorophyll, and TP-chlorophyll models. The Science Panel could run the national model to generate potential chlorophyll targets if the distributions are consistent. • The Science Panel could also compare TN and TP observations in Utah Lake with relationships in the national model. • Tetra Tech still has to complete S-R analyses on lake visitation and public perception. One challenge with conducting the S-R analysis for lake visitation is that there is no precise data on how many people are visiting Utah Lake; the only data available is on state park visitation, and that data is based on a statewide formula that calculates visitation based on the number of passes sold. The Utah Lake Authority will begin counting visitors in the future, but there is no past data on exact visitor counts. • DWQ is currently working with the primary investigators of the Utah Lake public survey, who surveyed the public to quantify the public perception of water quality. The primary investigators are completing that report, which should be available in late summer. • In the future, the Science Panel will need to discuss risk management and consider what prediction intervals and quantiles are appropriate for quantifying uncertainty. The high- yield approach may make that discussion less necessary. Science Panel Discussion on Outstanding S-R Analyses • It is appropriate to compare the site-specific model to the national model, but the Science Panel will have to assess the results carefully. 23 • One unique aspect of Utah Lake is its high turbidity. The light limitation studies on Utah Lake indicate that the average light intensity at a 3.2-meter depth is about 6% of the surface. This level of light might be limiting. It can be difficult to assess light limitation because of vertical mixing, which can bring cells into the photic zone. Science Panel Direction on Outstanding S-R Analyses Science Panel members supported having Tetra Tech compare the Utah Lake site-specific S-R models to the national models to see if they are consistent. PUBLIC COMMENT Members of the public commented on the Science Panel discussion at the meeting. Their comments are summarized below. • The Provo River Delta was recently completed. The Delta may potentially serve as a sink for nitrogen and phosphorus. • As June sucker populations recover, there will be an impact on zooplankton, which will, in turn, impact the Utah Lake. • The Timpanogos Special Service District (TSSD) hired researchers to conduct a Limnocorral Study on Utah Lake. The preliminary results from the Limnocorral Study suggest that if the sediment is stabilized by reducing carp and blocking wave action, the light goes to the bottom of the lake, producing filamentous algae and dropping nutrient concentrations. Additionally, zooplankton populations, particularly copepods and Daphnia, increase dramatically. Fish are excluded from the limnocorrals using nets and sandbags. Occasionally, fish will enter the limnocorral, at which point the researchers remove the fish by net. The researchers are testing various conditions using the limnocorrals, including adding nutrients. One potential way to stabilize sediments is to re-introduce mussels, clams, and macrophytes into the ecosystem. NEXT STEPS Tetra Tech will incorporate Science Panel feedback and continue to make progress on the S-R analyses. The Science Panel will continue to discuss the S-R analyses at their next in-person meeting in June.