Machine learning helps researchers categorize the ocean's ecology

Scientists at MIT used machine learning to find distinct points enabled them to split the world’s oceans into different “provinces” based on ecological makeup.

Scientists use chlorophyll seen in satellite images to define marine communities, MIT News reported. The amount of phytoplankton’s green pigment in an area can show how productive one ecosystem might be compared to another. What the amount of chlorophyll can’t do is show what plant and animal life is in that region, according to MIT News.

The scientists divided the oceans into more than 100 provinces.

“It’s like if you were to look at all the regions on land that don’t have a lot of biomass, that would include Antarctica and the Sahara, even though they have completely different ecological assemblages,” Maike Sonnewald, a former postdoc in MIT’s Department of Earth, Atmospheric and Planetary Sciences, told MIT News.

Researchers next grouped them under 12 categories called “megaprovinces.” From these 12 categories, researchers saw the regions had the same amount of life, but they had different balances of animal and plan species, MIT News reported. Those ecological subtleties enable researchers to track ocean health and productivity, Sonnewald told MIT News.

“Ecosystems are changing with climate change and the community structure needs to be monitored to understand knock on effects on fisheries and the ocean’s capacity to draw down carbon dioxide,” Sonnewald told MIT News.

The machine learning technique projects data from large complicated datasets into a simpler, lower-dimensional data set, which they named SAGE (Systematic Aggregated Eco-province method).

The researchers ran the algorithm on data from MIT’s Darwin Project, MIT reported. Humans couldn't work through the complicated, layer of data with 51 species of phytoplankton, including  the available nutrients, surrounding climate and how each species grows and interacts with each other, Sonnewald told MIT News.

“We started thinking about things like, how are groups of people distinguished from each other? How do we see how connected to each other we are? And we used this type of intuition to see if we could quantify how ecologically similar different provinces are,” Sonnewald told MIT News.

Researchers can use an online widget developed to find other similarities. They aren’t bound to the 12 categories Sonnewald’s team chose. 

Oceanographers can also use the tool to send their research ships into the right regions.

“Knowing what species assemblages are where, for things like ocean science and global fisheries, is really powerful,” Sonnewald told MIT News.