January 9, 2017

Dr. Don Anderson: Using robots to measure toxins in the ocean

Toxic algal blooms that infect shellfish and make people sick are happening more and more along the New England coast. Scientists haven’t had a way to predict where and when they will cause problems. But now MIT Sea Grant funded researchers from Woods Hole Oceanographic Institution, the Virginia Institute of Marine Science, and NOAA have developed a robotic system that they say can monitor blooms of toxic algae and their toxins in real time. This system, built around two instruments – a submersible “cytobot” that dives down to photograph algae where they emit their toxins and a second sophisticated robot that directly measures just how much toxin is around, is demonstrating a new vision for how we may protect seafood safety in the future.

Alexandrium fundyense, a species of algae found in the northeast U.S., as well as up and down the west coast, produces a neurotoxin that can contaminate shellfish beds. If the contaminated shellfish are eaten, people experience serious poisoning symptoms, and in severe cases, death. Current monitoring approaches rely on analysis of shellfish, a process that typically takes several days, a lag that potentially endangers seafood consumers. Now, researchers at Woods Hole Oceanographic have new tools that dramatically improve the quality of information available for managers and the public: robotic devices that monitor developing blooms and detect their toxins in real time.

Algae is bad for business

Every summer beaches on Cape Cod are filled to bursting with eager tourists looking to enjoy quaint towns filled with tasty seafood restaurants. Even a whisper of shellfish poisoning can have a huge impact on how many people buy seafood, or even choose Cape Cod for their summer vacation.

Poisoning cases from shellfish toxicity can have a long lasting ripple effect on the entire industry. Here in New England, because monitoring programs are purposely conservative, most poisoning cases from Alexandrium happen not from purchased seafood but when tourists, who don’t know any better, grab a few clams in an area that has been closed to shell fishing. But even one case is enough to make people hesitate or stop buying seafood altogether. It affects what fishermen can charge for the shellfish, even whether people buy seafood at all. NOAA reported shellfish sale losses of $23 million in Maine and Massachusetts alone from a red tide bloom in 2005.

It isn’t just people getting sick. When people get shellfish poisoning livelihoods are endangered.

Unseen until now

For the first time, this past March, in a small kettle pond in Nauset researchers were able to watch the Alexandrium bloom, grow and divide, in stop-frame photographs, while also monitoring toxicity levels, in real time.

“One of the reasons this last sampling effort was really exciting is we were able to use these great tools in a coordinated fashion, which provides us so much more information than if we deployed either one alone,” said Dr. Michael Brosnahan, a postdoctoral scientist in Dr. Don Anderson’s Harmful Algal Bloom Research lab at the Woods Hole Oceanographic Institution (WHOI).

Throughout the spring, from March until June, an Imaging FlowCytobot took pictures and identified individual plankton cells, while an Environmental Sample Processor measured the concentration of toxins in the water. The two instruments are connected to a platform that wirelessly transmits data to researchers. Anderson’s lab was able to monitor the entire bloom, as it happened, refining and tweaking instrument settings and positions throughout.

The Imaging FlowCytobot uses a focused laser beam to detect individual phytoplankton cells, then takes their picture at rates of up to 10 per second. The images are sent to a server at WHOI where a computer identifies the algae based on size, shape, and shading using a database of different types of algae organized into about 50 different training sets for the machine. But algae aren’t as cooperative as you might think, as cells grow and divide they look different. Some species like Alexandrium also have multiple life stages, each of which requires its own training set for accurate computer-based identification.

The Environmental Sampler Processer is a an automated molecular laboratory, able to concentrate particles in seawater, extract their toxins, then inject this extract into a chemiluminescent assay that is similar to a home pregnancy test. It provides data to researchers in near real-time that in the past was only available through sample analysis in a controlled lab setting. Researchers pre-load assays into a large cylindrical drum, load the 500lb instrument aboard a boat, then lower it down into the water where it is suspended from a winch that raises and lowers the instrument to where it will be most effective, until it runs out of assays. Because the instrument can only hold a limited number of assays, it is very easy to exhaust them while searching for where toxic Alexandrium might be in the water column. By simultaneously analyzing images from the Imaging FlowCytobot, the WHOI-based research team was able to make every one of the Environmental Sample Processor assays ‘count’, and more accurately interpret the relationship between Alexandrium abundance and water toxicity.

Researchers haven’t ever before had the ability to monitor these blooms out in their natural environment with this level of sophistication. The technologies integrated into the robotic platform are new to the world of oceanography, and every time they are used researchers are finding unexpected patterns and behaviors by marine organisms. On an applied level, having this type of real-time toxicity information enables researchers to alert state and local resource managers of toxic blooms that would have almost certainly caused serious health impacts. This exact scenario has occurred twice since the development of the robotic observatory. The first time, in 2012, the WHOI-based team detected a very early bloom of Alexandrium, so early that town officials hadn’t yet started their routine monitoring program for this species. The second time, in 2015, researchers monitored an intense bloom of a second species called Dinophysis. Working together with local, state and federal labs the team determined that high levels of the Dinophysis toxin had contaminated shellfish, leading to the first ever shellfishing closure due to this species in New England waters. Most importantly, in both instances, shellfish managers were able to act before anyone got sick.


More data means better predictions

Our harmful algal bloom problem will only grow in the future. Both rising ocean temperatures and nutrient runoff from growing coastal populations make blooms worse and we can expect to see more frequent blooms that last longer and increasingly pop up at unexpected places or times of year. Now is the time to get a handle on developing predictive models that incorporate the best possible characterization of how these toxic species behave in their natural environment.

For now, physically sampling shellfish is the best way we have to detect infected shellfish, but hopefully in the future an array of sensors will allow us to monitor larger areas more thoroughly while at the same time understanding how these species are adapting to the varied habitats along our coasts. That way we can do the best possible job protecting our seafood resources, even in places where these sensors aren’t.