CIRUN - Climate Information: Responding to User Needs

Chesapeake Bay Workshop

Integrating Climate and Environmental Information with Disease Surveillance to Address Pathogens and Algal Toxins of Concern to Public Health

Executive summary:

A CIRUN workshop, jointly sponsored by the NOAA Oceans and Human Health Initiative (OHHI) and the NOAA Climate Program Office, was held at ESSIC on February 21 & 22, 2012. Scientists from the ocean climate community and public health officials from State Departments of Health and Departments of Natural Resources in Delaware, Maryland, Virginia and Washington met for two days to discuss methods of providing reliable and actionable monitoring data, forecasts and models for climate and naturally occurring aquatic biological hazards to the public health community, with a focus on dangerous vibrio and harmful alga blooms (HABs) in the Chesapeake Bay. A companion regarding the Puget Sound workshop was held in Seattle on March 21& 22, 2012.

While these hazards currently pose a relatively small (but real) risk to human health, the advent of a major oyster aquaculture industry could significantly amplify that risk, as oysters form a well-known vibrio pathway to humans. This made the subject of the workshop a timely one for Maryland public health authorities.

Several key issues and potential actions emerged during the workshop. The immediate need to integrate physical, biological and public health data was recognized by all breakout groups.  Participants felt that such an integrated data system would make the combined results of observations and monitoring available to all and would allow future projects to focus more clearly on gaps in our knowledge. The ability to overlay physical climate data with vibrio and HABs data into a mapping system could assist in both industry planning and subsequent regulation of aquaculture in the Chesapeake Bay. The first step in such a project would be to develop a catalogue of the data that exists currently, which is spread across a wide set of diverse locations.  In recognizing the importance of integrating the data into one electronic data repository, the desire for increased monitoring and collection of additional data was also expressed.

A critical challenge recognized by the public health community was the absence of good dose-response data that would permit the translation of measured or predicted organism concentration into a measure of population risk.

The need to improve our understanding of how environmental factors affect the presence and growth of dangerous bacteria and HABs also emerged as an important and resolvable issue. Temperature, salinity, and precipitation changes, as well as nutrient loading to the Bay and land use/land cover changes were suggested as important factors associated with vibrio toxicity, but much work remains to be done to learn the thresholds at which point these factors may contribute to potentially serious levels of toxicity. That understanding would enable a more strategic design of future monitoring and data collection, and also support the ability to make longer range forecasts. Conducting combined retrospective analyses of vibrio cases in the Chesapeake Bay, the Puget Sound, and the Gulf of Mexico to determine to what degree environmental factors are coincident with the production of disease was suggested. 

With an integrated data system and a better understanding of significant environmental factors, the ability to produce reliable and useful forecasts for the public health community through a health early warning system becomes possible, not only on a daily to weekly basis, but also on monthly and seasonal time scales.  One of the challenges discussed in this context was whether the geographic and temporal scales of forecasts and models could be better reconciled with and useful for public health purposes, which are generally more immediate and localized. 

The importance of making public health authorities aware of the uncertainty in forecasts was recognized, and several pathways were suggested. A weekly health bulletin integrating information from relevant data sets relevant to HABs and vibrio (satellite, in-situ, and model runs) was the first suggestion. Educational workshops between local health departments in the Chesapeake Bay and the Puget Sound watersheds and climate scientists would initiate additional dialogue regarding available data sets, products and tools.

Plenary Session Report

Steve Halperin, CIRUN Director (Climate Information Responding to User Needs) welcomed everyone to the workshop. Antonio Busalacchi, Director of the Earth System Science Interdisciplinary Center ESSIC), also Chair of the University of Maryland Council on the Environment at College Park, provided an overall context to the workshop by noting that the workshop embodies what CIRUN and the new Council on the Environment are trying to do at Maryland - providing reliable and actionable climate information to stakeholders.  This workshop is being held in tandem with a workshop focusing on the Puget Sound. Internationally, under the World Meteorological Organization (WMO), there exists a global framework for Climate Services focused on food, energy, water and health. However, Busalacchi noted that many international meteorological organizations do not have an ocean hydrological component, which could present organizational challenges in going forward with human health initiatives. He also stated that the NOAA Center for Climate and Weather Prediction will be moving into the University of Maryland's Research Park, which will allow for continued collaboration. He thanked Dr. Chester Koblinsky, NOAA Climate Program Office (CPO), for providing workshop funding and recognizing the importance of such workshops. Busalacchi thanked the steering committee for their hard work in pulling the workshop together.

Koblinsky expressed pleasure with the wide diversity of participants at the workshop. He noted that NOAA is particularly interested in climate services. One of the biggest challenges is how to engage the people making decisions with those that are doing science. Thus NOAA is very receptive to CIRUN and UMD initiatives surrounding these workshops. Even though NOAA is not involved in the health care sector, they do want to know what climate information can be provided to that community.  Koblinsky noted that the oceans and marine fisheries are important to the NOAA Climate Project Office both locally and nationally, and thus the intersection with public health is important.

Julie Trtanj, Director of the NOAA Oceans and Human Health Initiative (OHHI) provided more context and background for the workshop. Currently, OHHI develops an early warning system which informs users of the risk from sea food consumption due to (HABs). Similar to WMO’s Global Framework for Climate Services (GFCS), the global observing system has agreed to share data on water born and water related diseases. In addition, state health departments are looking at prediction of vibrios and related health risks. As the steering committee began discussing the possibility of joint workshops focusing on the Chesapeake Bay and the Puget Sound, CIRUN and NOAA ideas were merged. The goal of these workshops would be to determine the environmental drivers of the bacteria and HABs to determine what can be delivered to the public health community.  As part of a larger effort the group was asked to consider not only what individual states need, but globally how vibrio and HABs changes can be tracked and the future risks associated with these changes.

1st Plenary Panel – Public Health objectives and needs Chair: Jerrod Davis

Jarrod Davis, Washington State Department of Health chaired the first session on public health objectives and needs. He noted that the panel would be discussing evaluating data that are human health hazards, setting policy and using data from forecasts. He also stated that when it comes to using predictive modeling to make decisions, it is extremely important to maintain credibility.

The first panelist of the session was Cliff Mitchell from the Maryland Department of Health and Mental Hygiene.  Mitchell noted the one overriding concern that the panelists hoped to address is how to promote and enhance the well-being of the state.  Each panelist bears some responsibility for this.  Mitchell deals mainly with food safety supply in a practical, hands on approach to problem solving. Food safety manifests itself in the tension between commercial industry and health. He is responsible for promoting the local food production of the Chesapeake Bay in a safe way. Mitchell stressed that agricultural harvesting from the Bay can suffer irreparable harm if the integrity of the food supply is questioned. For most food health outbreaks it is difficult to determine from where the shellfish came. Health officials cannot track the source and so the whole industry suffers.

Though the health department is often dealing with rare events such as marine toxins or infectious organisms, these events receive huge amounts of exposure, as was the case with instances this past summer of vibrio septicemia related to wounds. 

For health officials, appropriate communication of risk is crucial, as negative perceptions often make it into the press. Mitchell questioned whether a predictive model for HABs could provide information about the risk to humans since the number of cases of human HABs disease is very small in Maryland.  Though HABs is often not fatal for humans, he did note that there are animal deaths related to HABS disease. This example highlights the issue of how best to relay an increase in a relatively small risk in a way that will get people to take precautions. The risk needs to be relayed to commercial fishers as well as those that are enjoying the Chesapeake Bay and its tributaries recreationally.

David Blythe from the Maryland Department of Health and Mental Hygiene then spoke about infectious disease response from a public health practitioner perspective.  He questioned how climate and environmental information could be integrated into the daily practice of health officials. He noted that once there is an outbreak, an investigation occurs, and prevention communication is relayed to the public. Health care professionals and labs are required to report all cases to the state whether they are HABs, non-tuberculosis bacteria, or vibrio, which are all produced in the Chesapeake Bay. Surveillance of infectious disease then follows. The state investigates each report of illness and collects data including demographics, clinical characteristics of the illness, exposure information of the area, and exposure to shellfish and other fish. This information is then shared with the Food and Drug Administration (FDA) if shellfish were involved. In addition, wound and water exposure illnesses require survey information. If a common source is identified, the state becomes involved in procedures to stop ongoing transmission. There are currently not many outbreaks in Maryland related to the Chesapeake Bay, or caused by the consumption of shellfish. Forty-five cases were reported in 2010 as a result of the hot summer. (Illness appears to vary with the temperature.)  The most common shellfish outbreaks are neuro-related, and these prevention activities include public communication to the harvesters regarding harvest related policies, but also public communication through media alerts concerning raw shellfish consumption and the risk of exposure in water. Information is passed to health care providers on how to counsel high risk patients. There have been no water closures in Maryland because of infectious disease outbreaks; however, threshold vibrio levels which require intervention are still unknown. 

William Forlifer from the Dorchester County Health Department in Maryland provided a contribution to the discussion from a local perspective. Dorchester County has a very small population in Maryland, but a large geographic area. Several HABs and vibrio cases reported were reported in previous years.  In 2009 a pond was contaminated with HABs. The pond is about two miles long and feeds a river adjacent to the river that had a pfiesteria scare during that time. The DNR sampled the river and pond and discovered high levels of microcystis. Signs were posted, but the general public did not provide much of a response. In general, unless the outbreak or levels of contamination affects a recreation spot, a harvesting spot or public drinking water, most are not concerned. However, concern exists from health officials as to what affect vibrio have on the food supply since contaminated water is irrigating local crops. Forlifer also reemphasized the need for effective risk communication in getting people to respond to the risks.

During the discussion questions were raised on the relationship between vibrio and temperature. A strong correlation between the two appears to exist. The number of reported cases rises in the summer, as temperature and exposure increase. Questions still exist over a more precise way to identify outbreaks so that health officials can target specific areas. Climate scientists were curious if public health officials were observing any trends. The outbreaks are minimal and the specific denominators remain unknown presently. Specific aquaculture centers have been identified for monitoring but no real trend has been determined. Currently, oyster samples are taken twice a month in the Chesapeake Bay. Vibrio are more abundant in warm water, though there were no pathogenic strains found this past year.  When officials discuss illness data, it is difficult to confirm where exposure occurred. Food sources are not necessarily from Maryland waters. Although health alerts are useful to the public, the question was asked regarding what information would be useful from climate/weather forecasters.   Currently public health officials do not understand safe threshold levels of vibrio concentration and the relationship of environmental factors to vibrio levels. Risk numbers for vibrio do not exist and thus vibrio concentrations are not monitored at beaches or associated with wound related illnesses. Another question focused on whether or not enough data existed to draw baseline forecast. Do vibrio have variance factors for toxicity levels? And would a predictive model of pathogenic or toxigenic vibrio would be useful?

Effective communication between the DNR, State Environmental Departments, State Health agencies and the public was again stressed as key.  Scientists wondered if health officials could be given a reasonable probability on a certain time scale that conditions would change to increase illness risks, what time scale and change in trend would be useful.  The panel pointed out that for the food industry the emphasis would be more on aquaculture, and as the industry grows, food safety would be the priority.   It was stated that a better understanding about biology may be more important than plan a simple total vibrio count.  Currently, the vibrio management plan is state wide across Maryland and time and temperature requirements are uniform across the Chesapeake Bay with respect to harvesting.   It was suggested that focusing on time periods of 2-5 years and potential changes to temperature could help manage resources.

In summary, risk communication is reasonably effective, currently.   Climate predictions might be useful to aquaculture management regarding when and where to close growing beds, as industry is more reactive than proactive. These forecasts must be credible as closing a bed has a large economic effect and often officials are willing to take a risk rather than closing beds just due to high water temperatures.

2nd Plenary Panel – Environmental and health science perspective Chair: Juli Trtanj

The second plenary session focused on the environmental and health science perspectives. Trtanj noted that environmental drivers for each state are important, but encouraged participants to consider the larger context, such as the role of ocean acidification on vibrio. Not only are better tools necessary, but a better tracking system would also be advisable. A connection between environmental scientists and health officials is critical. She also noted that moving from surveillance to monitoring and prediction is important.

Glenn Morris, University of Florida, examines public health impacts within a research context to develop appropriate environmental models.  In South America, Haiti, and Bangladesh, water temperature and vibrio are well correlated and health concerns are associated with warmer water in these areas. Bangladesh does not have specific swings in water temperature and thus nutrient loading is critical there. Likewise, water temperature shifts in Haiti are minimal, and thus Morris is currently conducting experiments to determine environmental drivers on vibrio in that region. Careful monitoring is essential. HABs in the Virgin Islands are a major problem. Approximately 25-35% of the population is infected, but it is unknown if this is related to changes in water temperature due to global environmental changes. Data in the Virgin Islands suggest an inverse relationship. When water temperature increases, the number of incidents decreases. Moreover, a certain threshold temperature of 30C appears to exist at which point the organism appears unable to exist. The number of cases in Florida is rising due to rising temperatures because the temperature threshold has yet to be reached. Morris is working with the Woods Hole Oceanographic Institution on public health impacts and examining what are appropriate levels of toxins for certain species. Good models, good public health responses to identify outbreaks and diseases, and good environmental models with health parameters embedded in them are needed. 

Jay Grimes, University of Southern Mississippi, noted that environmental drivers such as salinity, turbidity, dissolved organic matter all influence cholera as well as chlorophyll A,  phaeophyton, and other harmful algae. In the Gulf Coast Lab they sample each of four different sites once every four weeks. His lab is developing molecular methods to detect vibrio, and is defining algorithms from remote sensing data to predict vibrio. A strong relationship exists between turbidity and primary productivity. Climate change and temperature increase appear to suggest an increase in vibrio. A recent publication on the North Sea demonstrated a relationship between temperature and vibrio and HABs. At 15C, these bacteria appear non-culturable, which would indicate a viable material threshold. Though bacteria may exhibit a dormant lifestyle at certain temperature thresholds, a lack of nutrients is also required for dormancy. The Gulf of Mexico exhibits huge temperature swings.  Even with water temperatures at 15C in the Gulf, bacteria are present in oysters. Though vibrio may become non-culturable in Gulf water oysters, they are still culturable in the sediment. Grimes noted that vibrio cause wound infections, while cholera leads to gastric issues. Those with liver impairments, immune deficiency or poor circulation are the most at risk. Currently, Grimes is finding abundant disease in Mississippi aquaculture. Crowded conditions lead to diseases.  No increase in fish diseases was observed in relation with the Gulf oil spill. Grimes believes that temperature alone can explain 50% of the disease risks, but by adding salinity information (remotely sensed data) scientists can explain 75% of the risks of disease in oysters. The task ahead is to figure out what parameters should be monitored to further explain the phytoplankton/HABs/disease relationship. Composition and virulence also needs to be examined as well as trends and changes in the environmental system.

Martella Du Preez from South Africa discussed an early warning system for cholera. Observation data in Africa are not reliable. Meanwhile, cholera is on the rise in Africa, but outbreaks are more sporadic in South Africa where water and sanitation are slightly better. Ground-truthing is essential in local areas, because otherwise modeling is useless.  DuPreez is examining such environmental drivers as rainfall, temperature, chemical and physical parameters, nutrients, and ultraviolet light to determine their effect on bacteria in water.  The question is how to provide stakeholders with useful information. The first step is providing awareness regarding cholera. In Africa political issues hinder the dissemination of information. The next step is showing stakeholders how to use the information in their decisions. Demonstrating how the software and input data can be used to create an outcome is critical. Her research group has not yet made a connection between cholera and inland waters.  A local dam with a water circulation driver seems to have eliminated toxic vibrio though other bacteria exist.
Andy Depaola, from the Food and Drug Administration (FDA), noted that there are few cases of vibrio illnesses in humans due to the consumption of raw oysters. However, climate is a major driver in risk and abundance of pathogens in shellfish. For aquaculture, the water temperature determines harvest time, while the air temperature determines the post-harvest process time. Salinity, turbidity, blooms, and natural disasters also affect the risk, but when/where and how the harvesting is completed is critical to the risk. He noted that the dose response is quite different among various types of vibrio. And often, the vibrio are from outside of U.S. boundaries, such as in the case of ships from South American ports introducing new vibrio to the U.S. (Mobile, AL). Regional and seasonal expansion of vibrio was also noted from migratory birds. The FDA needs additional information not currently captured on summaries such as harvest times and dates. Depaola stressed the need to create a database of harvest times and dates that could be correlated with weather conditions during harvesting. A new initiative, Vibrionet, is creating a risk reduction plan using the National Marine Fisheries Service (NMFS) landing data and illness data from the Center for Disease Control (CDC), though there is a one year lag in obtaining the data. A proposal was made to potentially link Vibrionet to the Ocean Observing System OOS). Predictions could be made based on climate conditions and harvesting information on the likelihood of getting ill. Risk mitigators could also be published such as closing an area for 1-2 weeks. Depaola thought FDA and Vibrionet would be great customers of the OOS.

Discussion focused on increases in the amount and intensity of precipitation and what contaminants it might bring.  Some of the environmental labs are looking at wet versus dry years globally, but extreme events and runoff would probably be a more important variable especially locally in the Chesapeake Bay. The Mississippi lab is examining the role of wind (speed/direction) as well. Salinity and nutrient loading with respect to runoff and in response to extreme events was stated to be equally important.  A question was raised regarding modeling turbidity in extreme rainfall and whether or not a relationship existed between the two. The answer is unknown presently. Turbidity and organic matter were examined in the Bangladesh studies. When organic matter is included in models to predict cholera, increases in infection are observed.

3rd Plenary Panel – Forecasting contributions Chair: Antonio Busalacchi

The final plenary session of the workshop focused on what forecasting techniques or systems could contribute to the public health sector. Busalacchi encouraged the panel members to comment on the current state of the science, ranging from forecasting (seasonal, annual, decadal…) to nowcasting, and to consider the linkages to the public health community.

Wayne Higgins of NOAA’s National Centers for Environmental Prediction (NCEP), saw a tremendous value in linking forecasts on all time scales to water.  NOAA is capable of monitoring and has a huge investment in climate services which the public health sector could use to their advantage. Health threats could be better predicted through strengthened a partnership with NOAA. He noted that common climate signals such as El Niño Southern Oscillation (ENSO) are important. When an El Niño occurs, the jet stream extends into the U.S and is connected with an increase in frequency and intensity of weather events which impact human health. Higgins also stated that pollens, changes in air quality and changes in pressure/humidity can affect human health (asthma, etc). These symptoms exhibit a strong connection between runoff, nutrient loading, and salinity in the Chesapeake Bay. One possible area of intersection would be to examine the connection to these climate variables to the amount of vibrio in species. However, as time scale increases, so does the uncertainty. The National Weather Service (NWS) issues operational forecasts and warnings such as hazard outlook products which cover the time scale of 1-2 weeks. These products are primarily used by the Federal Emergency Management Agency (FEMA). Higgins commented that drought in Texas impacts the Mississippi River runoff and outflow which relates to hypoxia.  He stressed that weather, water and climate information needs to be connected, and wondered how best to integrate the mutual capabilities of each community.  One possibility would be to develop integrated ecological forecasts to predict dissolved oxygen, nutrients and other health variables.

Frank Aikman, NOAA Coast Survey Development Laboratory, noted the limits and cautions for modeling, specifically the Chesapeake Bay Forecasting System (CBOFS). CBOFS has run 4-5 physical and biochemistry models to test for hypoxia in the Bay. Results indicate that the choice of physical model matters little. Hypoxia does not appear to be related to the vertical stratification of the model. CBOFS is now coupling with biochemistry models that can simulate and predict HABs. Several different approaches to the modeling are possible, but none are good without good observations. He noted many sources of uncertainties and errors which need to be provided to the user. Errors exist in the forcing, winds, heat flux, and boundary conditions as well as the model initialization. He reminded everyone that uncertainties also exist in the model output. 

Rick Stumpf, NOAA, has run an operational forecast for HABs for the past 7 years. He noted that the bloom in Lake Erie in 2011 would have covered the entire Chesapeake Bay. He cautioned that model output provides guidance, not a straight forecast. Data must be interpreted and currently data missing in the system includes the presence of toxin in toxic organisms. In addition, HABs is not well behaved and extremely difficult to predict. Furthermore, if the position of the bloom is unknown then the model will not work, which is a key problem in the system. He encouraged the development of an early warning forecast first, then a short term forecast and then a seasonal forecast. If a seasonal forecast can be done well, then public health officials can prepare and plan. Currently scientists try and tell people what they need rather than what health officials and managers actually need. He encouraged the workshop participants to continue to communicate their needs and abilities to one another.

Raymond Najjar, Pennsylvania State University, discussed the potential impacts of Anthropogenic Climate Change (ACC) on the Chesapeake Bay and whether or not ACC could affect restoration in the Chesapeake Bay. In a 2010 review article, he and others examined the sensitivity of the Bay to past climate activity and determined the Bay will likely function quite differently as a result of climate change. They found that intensification of the hydrological cycle will occur due to ACC, with increasing precipitation and streamflow in the winter and spring, which, in combination with higher temperatures, is likely to reduce bottom water dissolved oxygen levels and eelgrass coverage.  Other effects include a higher sea level, leading to loss of tidal wetland habitat, and warming and higher CO2 leading to an increase in HABs. The use of a more open-source modeling approach to address the Chesapeake Bay restoration has been recommended. He noted the response of the Bay to ACC is likely to be nonlinear.

During discussion, the type of useful forecasting was questioned. For example, is there a role for analog forecasting? Multi-model ensembles would be beneficial. Statistical downscaling rather than higher resolution models were also deemed useful.  Regarding analog forecasts, the lack of historical data would result in problematic forecasts, though persistence would be a good indicator for vibrio/HABs forecasts.  One possible solution would be to use the two week forecast from NOAA Climate Prediction Center (CPC) to run scenarios for pathogens.

Another question on validation and the role of public health data in modeling was raised. Is it more important to have more data versus validating a forecast with a subset of public health data? In order to forecast how many people are going to get sick, scientists need to know what information is needed. Many states are now using time and temperature periods to model vibrio. The FDA model just models the Gulf of Mexico because of a lack of data. Each coast is different regarding how vibrio behave. In order to build connections and have validated data, scientists need to know if the connections they are assuming are the correct ones. A suggestion was made to run hindcasts of the last five large toxic events in the Chesapeake Bay and mine past data to see what might be useful. A catalogue of vibrio cases might be useful in reanalysis case studies.  The OOS is now developing sensors that could detect bio-data, and advances in molecular technology will help.

Health officials cautioned that they still do not fully understand why organisms cause cholera or what actually causes the vibrio to create diseases in humans. Stumpf is running a vibrio model for the Gulf Mexico, but the mechanisms are still unclear and he would welcome feedback from the user community. Currently enough information for validation does not exist.

Charge to the Breakout Groups
Steve Halperin

Halperin reminded everyone that three different breakout groups with a combination of participants from the three different sectors (public health, environmental science, and forecasting) would discuss the same set of questions for the remainder of the workshop.  Each breakout group would create a report from their group which will be provided to participants in the Puget Sound workshop. Halperin encouraged the breakout groups to be clear and concise and to provide actionable items which can be achieved in the near term and over the next few years. The questions for the breakout groups were:

  • Given the state of science today, how should the results of monitoring and forecasts be provided to and used by public health authorities?
  • What are the most important steps to take to strengthen the value to public health authorities of monitoring and forecasts for vibrio and HABs? In particular,
    • How can monitoring, sampling and forecast methods be best integrated and made more effective to detect changes and provide early warning of changes and trends over time for the Bay.
    • How can these be harmonized with other efforts and regions, including specifically the Puget Sound?
    • How can the reliability of this information be quantified, so that the risks to public health can be effectively communicated?
  • Develop recommendations for a call for proposals and identify other actions that would accomplish some of the above. 

Breakout Discussions

Breakout Group 1
Chair: Ray Najjar
Day 1

General initial discussion:

The key outcome should be the communications aspect between the disciplines and between the users and scientists. Currently there is a disconnect between what decision makers want to hear and what scientists provide.  Uncertainties are huge in the medical and health community. The health community would like to use forecast models to know what the upcoming season will be like. It is not clear how uncertainty plays into all that. A major question is: what is the most relevant time scale? Is it seasonal?  For the public health community, the seasonal timescale would be good from a notification point of view. The environmental community might need longer timescales. All agreed that multiple species had been discussed, which may different forecasts or products. An additional question was how to communicate a risk that might not manifest itself for years. Each question listed could have different answers for the different species.

Responses to the three questions

Question 1: Given the state of science today, how should the results of monitoring and forecasts be provided to and used by public health authorities?

Current monitoring and forecasting systems were discussed. The industry monitors the full speciation of the algae community as well as the event response. Live counts are the fastest way. If levels are high, then follow up for toxins (rapid response monitoring) occurs. Currently algae are measured monthly, but it used to be bimonthly. The Department of Natural Resources (DNR) produces a map of different levels of HABs, but does not provide interpretation of the data. Additionally, there are no state regulations on any of these pathogens. The group determined that seasonal outlooks/forecasts would not result in a different reallocation of resources. Moreover, there are no resources for adaptive sampling at this point. The shellfish program monitors algae at fixed shellfish sampling stations (monthly) and does response samplings. Less toxicity monitoring is done, but fecal indicators and biweekly cholera are measured at 39 stations (through an ongoing research grant the past 11 years). Vibrio sampling in the Bay is monitored as well through the DNR. Chlorophyll remote sensing data exist (through SeaWifs, MODIS) but are not specific to HABs. Water quality from the Chesapeake Bay program is also monitored. In the European Union, but not in U.S., if vibrio hot spots are detected, then aquaculture monitoring in the area around them is begun to make sure shellfish remain healthy.   The U.S. is not even sure what levels to monitor for vibrio.

Regarding forecasts, there are 3 mechanistic models for 3 HABs species, and Jacobs has one for vibrio which predicts temperature/salinity distributions for where species could occur, there is a 3-day forecast, and 14-day forecasts have been provided at ESSIC/UMD (Murtugudde). The use of satellite information to fill in gaps from spot samples would be beneficial.  The Great Lakes algorithm for satellite data could possibly be applied for the Chesapeake Bay, for example by Stumpf’s group at NOAA. The question is how decision makers would use the forecasts. It was felt that forecasts would mainly be used to issue advisories. The technical information needs to speak to the uncertainties in the forecast.

The first question centered on vibrio, but uncertainties exist in how to use the data. HABs forecast systems are much further along, and already have established threshold levels. The FDA bases most of their information on LD50s, but the public has no concept of what this means. For HABs, the suggestion was to integrate data from models, monitoring, and satellites and provide the information to public health officials. Data should also be regularly reported to the agencies.

Question 2:  What are the most important steps to take to strengthen the value of public health authorities of monitoring and forecasts for vibrio and HABs? In particular,

  • How can monitoring, sampling and forecast methods be best integrated and made more effective to detect changes and provide early warning of changes and trends over time for the Bay?
  • How can these be harmonized with other efforts and regions, including specifically the Puget Sound?
  • How can the reliability of this information be quantified, so that the risks to public health can be effectively communicated?

All agreed that better monitoring is essential. It needs to be more comprehensive in time and space. Currently vibrio information is not being shared at all, because many are not sure what to do with it.  A suggestion was made to refocus the monitoring of vibrio on certain pathogenic strains that would be harmful to humans.  Basic science still needs to be done, especially as it relates to food security, however, better methods for utilizing the models needs to occur as well. In the Chesapeake Bay we need to better understand the factors that influence vibrio and marine algae. The technology exists to do the necessary modeling and forecasting, but implementation will require financial resources, either new or refocused from existing resources. Monitoring resources for HABs is dwindling (CDC money going away). Suggestions for better communication included providing hot spot maps to the public and public health newsletters. In addition adaptive sampling (integration of monitoring/modeling to focus sampling) should be done.

Question 3:  Develop recommendations for a call for proposals and identify other actions that would accomplish some of the above. Are there ways to bring the three groups represented together?
The group’s list of recommendations included:

  • Understanding vibrio thresholds.
  • Rapid and direct detection of toxins.
  • Develop dose response functions.
  • Public/private partnership development (aquaculture) nationally

Other general discussion:

Vibrio are a high priority because they can shut down the Chesapeake Bay oyster production, which would have a huge impact on the food industry. Better data is needed, and the ability to put these data into a mapping system could assist in regulating aquaculture. For food protection, the post-harvest processing is the most important. Harvesters need to know the time and temperature requirements to kill bacteria. If more aquaculture occurs in the Bay, then post harvesting processing regulations will have to be put into place.

Ultimately, physical, biological and public health data must be integrated.  The current 3-day forecast is just a management tool for making warnings or shutting down locations, but it would be nice to get to the point of notifying the medical community of certain conditions to raise awareness to support the diagnosis of illnesses.

Day 2

What we know and what we need to know:

Some models could predict hot spots at places where we don’t normally sample and for this we need to know that the spatial data of reported illness is valid.

Underreporting occurs more frequently with minor/mild cases of illness and varies by location as well. Reporting also depends on how much testing was done, how many care facilities exist, and the socio-economics of the region. Health officials rarely take time to follow up and figure out where the illness occurred.

If there is a drought, then high numbers of bacteria are possible in a coastal population. The question of whether or not you can go from fingerprint to the dose response to the effect was asked. The response was that it is possible but not presently done.  A possible goal could be to go from the fingerprint to the dose response; however, hazardous dose levels are currently unknown.

Recommendations that can be accomplished with no additional resources:

  • Communication with Public Health authorities: Bulletin of all data produced weekly – satellite, HABs data, sample data, environmental data.
  • Educational workshops to inform local health departments of potential vibrio risks; and what data are available.
  • Immediate development  of an MOU between agencies (state Public Health Departments, NOAA, FDA, State DNR) to coordinate data transfer and data integration.

Recommendations that can be accomplished in 1-3 years:

  • Focus resources/objectives/goals in 2-3 years to have a dose response for the pathogenic strains identified from clinical samples in the Chesapeake Bay.
  • Hindcast analysis linking clinical signals to environmental variables (temperature salinity, precipitation); this will require obtaining the data from the State.
  • Determination of which environmental conditions favor the production of vibrio disease.   This will require retrospective analysis of past cases and pattern matching of vibrio data with environmental data.  Reporting of human cases in and improved vibrio surveillance will be necessary.
  • A 3-day to seasonal forecast system, with initial steps toward validation, to warn the aquaculture industry, public health officials, and recreational users about conditions likely to increase risk.
  • Development of a dinophysis model tied in to the Puget Sound, and a review of current models for microcystis.

Long term goals:

  • Regulators would like to know the variability of fingerprints/markers of toxins in the Puget Sound and the Chesapeake Bay (2-3 year), with a link of human (clinical) isolates to environmental source.   This may require that the fingerprints be site specific.
  • Diagnostic testing for HABs is needed.  This requires identifying biological markers of exposure. For the Chesapeake Bay the initial focus should be on 2 markers (microcystis, and possibly dinophysis); over the longer term, other blue/greens could be added.
  • Given known climatic changes to the Chesapeake Bay, develop the ability to inform industry where to farm and from where to steer clear.
  • Determine virulence factors (e.g. genetic code) that cause disease for vibrio.  (Different factors exist for each species.)

Breakout Group 2
Chair: Tom Miller
Day 1

Question 1: Given the state of science today, how should the results of monitoring and forecasts be provided to and used by public health authorities?
Additional questions produced by the breakout group included:

  • What are we trying to work towards? 
  • In what ways are things inadequate today?

Surveillance data is needed and a good statistical probability model with concrete data to support it. Question:  Is there a need to monitor as opposed to simply responding to outbreaks?  Outbreaks are unknown if monitoring (epidemiologically) is not occurring. Additionally, without baseline data there is nothing with which to compare outbreak data. In Washington State, environmental samples are collected to do predictive modeling, but how can we build upon that to get to the next level?

Another question dealt with the lack of optimal conditions for today’s known pathogens. These need to be known for the necessary monitoring that so that scientists can develop tools to track emerging threats and/or multiple pathogens.   A third issue is the widespread underreporting of illness. (An earlier CDC report stated that vibrio illness is underreported by a factor of 20; this was later updated in the new CDC report estimating that only 1 in every 158 vibrio illnesses is reported.)

It is extremely difficult to go through COVIS (Cholera and Other Vibrio Illness Surveillance) to get data to pair up with modeling efforts in order to validate predictive efforts. Health monitoring system improvements that need to be made include: doctors asking the right questions to pinpoint infection sources, since current data leaves many questions about where the exposure took place. A recommendation was made to build networks of outbreak updates/advisory lists as an informal network of communication to keep public health managers and researchers on the same page, as well as to update research efforts on an informal basis towards optimizing research efforts (especially those related to virulence).
Since Washington State reports that it takes 10-14 days after an illness to get information on it, how far ahead would an environmental monitoring program need to produce data for the public health managers to be able to use it? Real time data indicating the risk over the next couple of weeks would be optimal.  In Maryland, conditionally approved areas are the basis for closure/opening. When a 24-hour accumulation of rain events occurs in these growing areas, the growing area is typically closed for 3 days. There is value in knowing predicted seasonal conditions (e.g.  wet/warm/dry) for growers, but would there be a planning element or just a management effect from these seasonal forecasts?

Other questions from the scientists to the public health community included:

  • Is there some sort of environmental monitoring that would cause the public health managers to do things differently?
  • If you close an area, when do you open it? Under what environmental conditions?

Regarding ocean acidification, large PH change is not required to break down an oyster shell. There is a great deal of tension between starting a booming aquaculture industry and the unknown risk of illness.  There is a need for a temporal map of data that people are collecting and to overlay many different parameters on it.

Question 2:  What are the most important steps to take to strengthen the value to public health authorities of monitoring and forecasts for vibrio and HABs? In particular,

  • How can monitoring, sampling and forecast methods be best integrated and made more effective to detect changes and provide early warning of changes and trends over time for the Bay?
  • How can these be harmonized with other efforts and regions, including specifically the Puget Sound?
  • How can the reliability of this information be quantified, so that the risks to public health can be effectively communicated?

Question 3:  Develop recommendations for a call for proposals and identify other actions that would accomplish some of the above. Are there ways to bring the three groups represented together?

Response to the above two questions was to try to integrate vibrio and water quality plankton data, although it is unclear how to link these from a spatial standpoint. The sampling design at present in the Chesapeake Bay gets at the largest spatial coverage. It covers phytoplankton, zooplankton, bacteria, CTD (conductivity, temperature, depth) casts, and nutrient analysis. A suggestion was made to overlay the data and see if there are relationships that exist.  It might be possible to overlay the data with the hindcast information to integrate the data and models.

Current models running:

  • CBOFS, an operational model, is run out of NOAA every 6 hours.
  • ChesROMs is run out of ESSIC, is more experimental, and has a biogeochemical component (CBEPs).
  • Tracy McGrady has a version of ROMs for the Puget Sound.
  • Pannel is a model for the Puget Sound.

What could be added to the models to give them more utility?

  • Jacob’s group has recently added zooplankton.
  • Adding alginolyticus to the sampling.
  • Validate the cholera model with the current database.
  • Validate the models with illness data from CDC.
  • It may be better to get illness data at the state or local levels.

There is a great need to facilitate communication among health officials and between the states. An important consequence would be improved training of health practitioners /doctors to ask the right questions of those infected. One way to accomplish this could be through an email list-serve, or possibly webinars. The group encouraged everyone to take up Wayne Higgins on his offer to expand his models to the mid-Atlantic. In addition, the assembling of data from various researchers at a common location was deemed crucial. ESP sensors to monitor for multiple pathogens should be included.

Day 2

What we know and what we need to know:

A better understanding of what makes these various strains virulent is needed. Vibrionet is a data integration piece that can be populated with data and would be useful in this regard. A goal of creating an integrated data system /local/national/global that could inform the Vibrionet, perhaps under the framework of the International Ocean Observing System (IOOS), would be beneficial. Species presence/absence and abundance data would also be useful. Currently within the IOOS there is no national strategy for biological data, but that group could be encouraged to pursue this sort of agenda. Davis is giving data to Depaola that includes clinical information on illness and environmental data.

MLST (multi locus sequence typing) data should be paired with CDC data so that data is not lost when CDC access goes away within a year. Washington State has a powerful dataset of clinical cases, harvest sites, environmental and climate data to match those sites and dates of harvest. Landing data is also important to understand why the number cases go up or down. In Washington, the landings of oysters are coded by different color tags based on how the oysters can be used (shucking only, half shell, etc.) Such data from Maryland can be used in the pilot project that FDA is conducting right now and can be submitted in excel format. A large dataset based upon individual questions about ecosystem health can be useful to all involved including other parties that are not necessarily involved now such as the Environmental Protection Agency (EPA), and United States Department of Agriculture (USDA).

Additional information for public health managers to make decisions based on science:

Integration of the large datasets of vibrio data from Maryland and Washington is important. Testing the Regional Ocean Model System (ROMS) models that are available in both areas to see if there is overlap would be useful as well. ROMS may be able to handle the HABs scenarios in the Puget Sound. In the Chesapeake Bay, the situation may be a little more difficult because bloom initiation causes are unknown.

Putting forward a health message (perhaps a month ahead of time – a seasonal message) based on forecasts could be useful, but the confidence of the forecast needs to be relayed as well.  From that confidence level, a public health forecast could be issued.  The Weather Service is actively trying to engage public health issues. For HABs detection, rapid, quantifiable methods for in-situ and lab-based measurements of microcystis are needed. Sorting out the virulence is most important with vibrio, followed by rapid tests to detect those strains, where possible in situ.

Breakout Group 3
Chair: Don Milton

Day 1

Question 1: Given the state of science today, how should the results of monitoring and forecasts be provided to and used by public health authorities?

Question 2:  What are the most important steps to take to strengthen the value of public health authorities of monitoring and forecasts for vibrio and HABs? In particular,

  • How can monitoring, sampling and forecast methods be best integrated and made more effective to detect changes and provide early warning of changes and trends over time for the Bay?
  • How can these be harmonized with other efforts and regions, including specifically the Puget Sound?
  • How can the reliability of this information be quantified, so that the risks to public health can be effectively communicated?

The most important steps are to detect change, develop an early warning system, and communicate quantities and uncertainties to the public. Uncertainty is a large issue especially regarding what toxicity levels are acceptable. Standards for contaminated water need to be established, but health officials are still unsure how to approach establishing such standards.  Probability distributions of risk would be helpful. Vibrio is extremely diverse and the variance factor is unknown. Cases of vibrio are not high. Most cases don’t have efficient dose, and it is not easy to predict the risk based on the population.

There are certain biological questions that must be answered such as the effect of environmental change on vibrio and HABs populations. Comparing environmental data with reports of vibrio and HABs population would be very useful. Xin-Zhong Liang can determine what factors of temperature and precipitation are important.  Predictive models should be based on the understanding of vibrio. Since vibrio cannot be rigorously tested in a laboratory, would it be possible to combine field measurements along with environmental condition predictions to allow process understanding and model development? Environmental models are ready to run at ~10 km resolution. If a specific area for a potential vibrio outbreak were identified, the model could be applied. A question was asked if hindcasts of vibrio outbreaks are useful for retrospective modeling studies. This could be useful if NOAA were to develop an integrated biological prediction system.

Sampling HABs in the Chesapeake Bay has occurred for the last two years. The sample size for three coastal studies includes biweekly spring and summer measurements of the bloom, as well as environmental condition changes. A list of the most likely environmental factors to link with vibrio bloom would be a starting point for environmental modelers to begin developing predictive models.

The question of significant increase in pathogenic bacteria potentially being an issue was raised. That possibility does exist as a result of climate change. Liang could estimate the probability over the next 50 years of such a significant increase. Probability distributions for population in the Chesapeake Bay over next 20 years could then be produced. The environmental factors that go into the modeling are very important. A large amount of surveillance must occur as physical, biological, and ecological data must all be factored into the model. Once results determine the conditions for vibrio to flourish, a genetic analysis of pathogenesis could be created.  For HABs the environmental drivers of importance include salinity, enrichment of nitrogen, temperature, and runoff.

Question 3:  Develop recommendations for a call for proposals and identify other actions that would accomplish some of the above. Are there ways to bring the three groups represented together?

  • Interdisciplinary collaboration.
  • Anthropogenic induced acidification.
  • Need to look at HABs and vibrio together.

There are four clusters to focus on for recommendations:

  • Data collection – which groups are collecting data?  Ensure there is a common place where all data is stored for public use.
  • Develop a detailed process model, with prediction accuracy.
  • Develop estimates of elements which can help focus future model refinement.
  • Start the production of an early warning system.  The first versions of the system may be quite crude and therefore should not be distributed outside of the public health community.  

Locations of hotspots of HABs are known for the Chesapeake Bay, usually in short periods, and seasonal, with monitoring since 1983. Similar monitoring data are available for vibrio in the Puget Sound.   An effective early warning system will require better data, and so monitoring system must be designed to focus on high priority areas. In terms of risk assessment, models should predict the probability of illness to warn people. One predictor could be where the toxic organisms are going to appear.  The real question is the threshold level at which the toxic organisms pose a risk. Since the number of cases in the Chesapeake Bay seems to be very low, projections of 10-20 years could be useful, but actionable predictions of the probability of exposure could be a problem, especially as the pathways are unknown. Focusing on the hot spots for HABs instead of vibrio might be better. The criteria for an early warning system should include:

  • Is this going to identify a serious danger?
  • Are fishing industries significantly affected?
  • Does the system identify a risk of public danger?

Day 2

What we know and what we need to know…

There is a need to move to a sequential string data system in the future, and someone needs to re-run the archives.  Data needs to be sorted into reliability groups which show variations from the sequence data.

For forecasting, we need to know the key environmental factors for vibrio growth as population increases, intensity of storms increases, and nutrients increase.  Temperature extremes and distribution are equally as important. During extreme events, heavy rains transfer large amounts of heat and nutrients to the Chesapeake Bay. These are environmental factors which appear to be   associated with the growth of vibrio populations.  Additionally, storms affect the salinity and turbidity which may also may also relate to vibrio population growth. 

The database required for a predictive model must be identified.  It should include environmental factors such as rainfall, land use/land cover, and nutrients. This kind of data is already being collected. The ecological data necessary is unclear. Currently only chemical compositions and very basic variables are measured.  The following data and specimen sources were suggested (chiefly from Maryland state agencies and universities):

  • Specimen libraries:
    • Maryland Department of the Environment (MDE) prospection fecal coliform
    • Academic collections
    • DHMH isolates
  • Data sources:                                                  
    • Department of Natural Resources (DNR)


  • Department of Mental Health and Hygiene (DHMH)
  • University Researchers
  • National Marine Fisheries (consumption)
  • National Socio-Environmental Synthesis Center

Two methodology schemes were proposed to establish an early warning system as outlined in the chart below:


Recommendations that can be accomplished in 1-3 years:

  • Understand ecology of HABs.
  • Build a catalogue of available data.
  • Learn what the relevant external factors are.
  • Design improved and coordinated monitoring systems and interpolation of resulting data.


Plenary Reports from the Breakout Groups

Breakout Group 1

  1. Recommendations for immediate action, which require no (or very modest) additional resources:
    1. Develop a Memorandum of Understanding (MOU) among agencies (MD, VA, and DE state public health departments; NOAA; FDA; state DNRs; etc.) and other interested parties (e.g., universities) to coordinate data transfer and data integration.  Such an MOU will be essential for setting a tone of cooperation between interested parties in making progress on improving public health outcomes related to HABs and vibrio in the Chesapeake Bay.
    2. Produce a weekly public health bulletin that integrates information from relevant data sets and models relevant to HABs and vibrio in the Chesapeake Bay.  This would include monitoring data, satellite products, and forecasts
    3. Hold an educational workshop for local (e.g., county and municipal) health departments about vibrio in the Chesapeake Bay.  The workshop would initiate dialogue and inform local health departments about the state of the science (including data availability) and the potential vibrio risks.
  2. Recommendations for actions over the next 1-3 years, which would require modest additional resources:
    1. Develop dose-response functions for pathogenic vibrio strains identified from clinical samples originating from the Chesapeake Bay.
    2. Improve follow-up and reporting through interviews and surveys after the occurrence of illness likely caused by vibrio in the Chesapeake Bay region in order to better identify conditions and environmental factors leading to illness.
    3. Conduct a retrospective analysis of vibrio cases in the Chesapeake Bay region in order to better determine the environmental conditions (e.g., temperature, salinity, and precipitation) that are coincident with the production of disease caused by vibrio.
    4. Develop a dinophysis model for the Chesapeake Bay and review current Chesapeake Bay models for microcystis.
    5. Begin development of a forecast system covering the three-day to seasonal time span for vibrio and HABs.  The development of the system would include continuous model evaluation and refinement.  The forecasts could be used by public health officials to develop warnings and by shellfish producers and state DNRs to focus and enhance monitoring.
  3. Recommendations for actions over the next 3-10 years, which would require substantial additional resources:
    1. Develop a clinical diagnostic test for Chesapeake Bay HABs.  Currently, it is not possible to determine unequivocally if an illness is caused by a specific HABs species.  Such a capability is needed.  The focus should initially be on microcystis, with subsequent work on dinophysis and other blue-green HABs species.
    2. Determine the genetic virulence factors that cause disease by vibrio.

Breakout Group 2

Relevant scales for forecasts:

  • The workgroup discussed the temporal and spatial scales of response of management systems.  Currently both MD and WA can respond within 12 hours of adverse weather or disease reports for individual shellfish grounds, suggesting these set the minimum scales for prediction.   Forecasts at longer and broader scales could provide a basis for “priming the pumps” that may include initiating empirical sampling programs or stakeholder educational initiatives.  The workgroup also noted that forecasts can play a role after HABs or pathogen outbreaks by shortening periods of closure of shellfish grounds, or lack of recreational access.
  • The workgroup noted that the current empirical platform for forecasting is spatially and temporally coarse.  Continuous monitoring of physical variables has changed our understanding of hypoxia and may produce similar impacts for HABs and pathogenic organisms.  This would require development of the next generation of sensor technologies.

Pre-requisites for forecasts:

  • We need to quantify dose-response relationships for both HABs and pathogens.  Research needs to focus on both human dose/exposure – health response and environmental dose – health response levels.  Central issues involve identification of critical abundance thresholds.
  • The workgroup noted that dose-response relationships need not be of the classical continuous form, but could be probabilistic non-parametric models.
  • Importantly, the workgroup noted that if such relationships cannot be developed, then negative health outcomes are effectively stochastic and forecasting has little benefit. This would force agencies into a mode where they respond to outbreaks but cannot prepare for them.


  • Immediate -
    • Data assembly.  The workgroup recommends that immediate steps be taken to assemble existing, spatially-resolved environmental and human-health data to develop a GIS database that provides a platform for statistical analysis to identify patterns and relationships.  Initial analyses should focus on hotspots of disease.  The goal is to develop dose-response relationships by overlaying existing spatially-resolved information.
    • Methods documentation and calibration.   Agencies use different laboratory methods to quantify HABs and pathogenic organisms.  It is important to document, and where possible, to inter-calibrate these methods to provide a standardized database for analysis.
    • Integrate existing HABs and pathogenic organism sampling with on-going water quality sampling programs.  For example in the Chesapeake Bay, water quality sampling and HABs/pathogen sampling occur at different places and on different schedules.  Co-sampling will provide benefits to both programs.
  • Minimal additional funding -
      • Enhance data collection / integration by promoting electronic collection/storage of human health and environmental data.  Currently human health data is collected and entered manually to meet separate reporting requirements.  An integrated, single point of entry data program for human health data would provide a database that could produce formatted reports to meet agency requirements.  Similarly, benefits would accrue from integrating environmental and genetic data into an integrated database.  Data integration among multiple user groups allows synthetic meta-analyses that may produce important new insights.
      • Support data exploration workshops to develop testable, well-posed hypotheses that the can be the focus of more sophisticated data analyses.
      • Support expanding sampling programs that focus on empirically quantifying occurrence, distribution and abundance of pathogenic microbes, in particularly focusing on the occurrence of toxic forms.
  •  Enhanced funding -
  • Genomics.  A revolution in the availability of low-cost genomics data is underway.  These new sources of data will likely provide insights into drivers of virulence of pathogenic organisms.  These might include:
      • Incorporation of novel genetic material from the environment into pathogens.
      • Influence of the human microbiome into the environment as it affects the evolution of infectious disease.
  • Rapid quantification methods for detecting environmental presence of HABs and pathogenic organisms.  New sensor technology has the potential to revolutionize our understanding of the occurrence, distribution and abundance of HABs and pathogenic organisms.  Sensor technologies that identify toxic strains should be emphasized.
  • Design, implementation and analysis of a continuous monitoring network for HABs and pathogenic organisms.

Breakout Group 3

  • As the aquaculture industry begins to rapidly develop in the Chesapeake Bay, reliable climate projections for the Bay are needed. A repository of data must be built at the scale of ecology and operation of the Bay. Data must be integrated from state agencies, federal agencies and universities with a standardization of metadata tags.  Improved and coordinated monitoring systems and interpolation of resulting data should be designed.

Back to Top