While most species of algae are harmless to humans and animals, some—like the Karenia brevis organism that causes red tide in the Gulf of Mexico or Alexandrium spp. found throughout US coastal waters—are toxic. When these species multiply and create harmful algal blooms (HABs), they can wreak havoc on human and marine animal health, contaminate seafood, and devastate local economies.
Red tide can be especially harmful to asthmatics and people with other chronic respiratory diseases, and these groups are more likely to need emergency medical attention during blooms. One Sarasota, Florida-based study estimated that the cost for these increased ER visits ranged from $500,000 to $4 million, depending on a bloom’s severity and duration.
Scientists and health experts have developed tools designed to better protect the public from the health impacts of toxic blooms, including triggers for the closure of shellfish beds, weather forecasts that predict red tide impacts on coastal communities [https://www.weather.gov/mdl/ndfd_maps], and a reporting system that informs the public when red tide is impacting beaches [https://visitbeaches.org/].
A fully assembled HABscope 2.0tm with a flask of water containing Karenia brevis. (Image credit: GCOOS)
The latest tool developed to protect public health is the Red Tide Respiratory Forecast [www.RedTideForecast.com], which was created thanks to a new ocean observing tool developed by the Gulf of Mexico Coastal Ocean Observing System (GCOOS) and Texas A&M University, where GCOOS is based.
“Red tide impacts can be really variable because of wind patterns,” said Dr. Barbara Kirkpatrick, GCOOS Senior Advisor and an environmental health scientist who conducted the first studies documenting the impacts of Florida red tide blooms on human health. “There are very few days when all beaches will be affected by red tide, and often your favorite beach is only affected for part of the day. So, we needed a Red Tide Respiratory Forecast to allow people to see which beaches might be impacted by red tide and at what time of the day. This would allow them to use the Respiratory Forecast the same way they use other weather reports to plan beach walks and other outdoor activities accordingly.”
Predicting Toxic Blooms
To power the forecast, GCOOS Research Specialist and Product Developer Robert Currier developed HABscope 2.0tm, a phytoplankton classification system that not only identifies toxic species but also provides information on concentrations, allowing modelers to more accurately predict toxic bloom movements and where impacts might occur.
While several commercial phytoplankton classification instruments are available, they cost well over $100,000 per unit, require highly skilled technicians for operation and maintenance, and still rely on scientists to analyze the imagery they gather. HABscope, on the other hand, is built out of a commercially available microscope and 3-D printed case with a single board computer and camera. It costs just $500 per unit.
This screenshot shows how the AI classified a sample image from HABscope 2.0tm. The green squares were identified as Karenia brevis, and the red squares were marked as non-Karenia. (Image credit: GCOOS)
But the real beauty of the HABscope tool lies in its ability to use Artificial Intelligence (AI) to recognize HAB species and count individual cells—automating a labor-intensive and repetitive task that had required highly skilled technicians looking at water samples through a microscope to identify and count each individual organism. Kirkpatrick says, “Counting microorganisms like Karenia or Alexandrium is a time-consuming process, and it requires a significant amount of training. Instead, Bob was able to automate the process, which allows us to use volunteers with some basic training.”
The Nuts & Bolts of HABscope
To develop HABscope, Currier used the open-source machine-learning software TensorFlow to build a model that would help the computer automatically identify and count Karenia brevis. Now, when a volunteer collects a water sample, they place it on a slide and use the microscope adaptor to take a short video. The video is uploaded to Digital Ocean, where the AI takes over —recognizing and counting individual cells to determine whether Karenia is present and at what level. Reading the slide takes the AI about 10 seconds, compared to 10 to 15 minutes it takes a trained human.
That information is then fed into a model developed by the National Centers for Coastal Ocean Science (NCCOS) of the National Oceanic and Atmospheric Administration (NOAA) that combines cell concentrations with wind patterns and information on ocean currents to develop the Red Tide Respiratory Forecast, which is hosted on the GCOOS’s ocean observing data portal and is part of the NOAA-NCCOS nationwide Harmful Algal Bloom Forecasting System.
The components of HABscope 2.0 tm are ready for assembly. (Image credit: GCOOS)
“In all, it took some 58,819 images of K. brevis cells to develop the model and ‘teach’ the computer to recognize red tide,” Currier said. He has since trained the system to also recognize Alexandrium spp. and Pyrodinium—toxic algae species that can harm human health. In addition to the Gulf of Mexico, the tool is being used by the Alutiiq Pride Marine Institute in Seward, Alaska, and the Virginia Institute of Marine Science.
A study published in the peer-review journal PLoS ONE showed the effectiveness of using citizen science volunteers—the team includes about 50 volunteers—working with HABscope to help increase the size and accuracy of the red tide monitoring network to better protect public health from the impacts of toxic algae in the Gulf of Mexico.
GCOOS Volunteer Coordinator, teaches Emily Colson of the University of Florida’s Nature Coast Biological Station on Cedar Key, Florida, how to use HABscope 2.0tm. (Image credit: GCOOS)
The study—“HABscope: A tool for use by citizen scientists to facilitate early warning of respiratory irritation caused by toxic blooms of Karenia brevis”—showed that it’s possible to:
- Increase red tide monitoring over a wider geographic area;
- Provide more accurate scientific data about red tide concentrations based on water samples instead of
- anecdotal reports from beach observers;
- Expand the red tide monitoring network throughout the Gulf states in a cost-effective fashion.
The Red Tide Respiratory Forecast was initially developed by the National Oceanic and Atmospheric Administration's (NOAA) National Centers for Coastal Ocean Science in partnership with GCOOS, the Florida Fish and Wildlife Conservation Commission-Fish and Wildlife Research Institute (FWC-FWRI), and Pinellas County Environmental Management. Additional partners now include Texas Sea Grant, the Sanibel-Captiva Conservation Foundation and Pinellas, Lee, and Collier (Florida) counties, and the Florida Department of Health’s Sarasota office. Initial development was funded through the NASA Health and Air Quality Program. Additional funding has been provided through the US Integrated Ocean Observing System and the NOAA-NCCOS multi-year “Monitoring and Event Response for Harmful Algal Bloom (MERHAB)” program as part of a nationwide effort to improve monitoring of and response to harmful algal blooms (HABs) along US coasts.
To find out more about GCOOS, visit: www.GCOOS.org
This feature appeared in Environment, Coastal & Offshore (ECO) Magazine's 2023 Deep Dive I special edition Ocean Observation, to read more access the magazine here.
