Artificial intelligence reveals how susceptible eels are to warming and disease

Artificial intelligence reveals how susceptible eels are to warming and disease

Meadow eels are photographed at low tide at Falls Bay Biological Preserve, Washington. Grassland is a vital coastal type of seagrass for fish habitats, biodiversity, beach protection and carbon sequestration. Credit: Olivia Graham

A combination of sophisticated environmental field methods and artificial intelligence have helped a multidisciplinary research group discover eel wasting disease at nearly thirty sites along 1,700 miles of the West Coast, from San Diego to southern Alaska.

Key finding: Seaweed wasting—caused by the organism Labyrinthula zosterae detectable by pests on blades of grass, confirmed by molecular diagnostics—is associated with warmer-than-normal water temperatures, especially in early summer, regardless of region. Grassland is a vital coastal type of seagrass for fish habitats, biodiversity, beach protection and carbon sequestration.

A Cornell research team — led by Carla Gomez, Ronald C. and Antonia V. Nielsen, Professor of Computing and Information Sciences in Cornell’s Ann S. Powers College of Computing and Information Sciences, and Drew Harvell, Professor Emeritus in the Department of Ecology and Evolutionary Biology (College of Agriculture and Life Sciences; College of Arts and Sciences) — reported for their findings on May 27 in Lake science and oceanography.

Co-lead authors are Brendan Rapazzo, a computer science doctoral student, and Lillian Aoki, a former postdoctoral researcher in the Harvell Laboratory, now a research scientist at the University of Oregon. ecology and evolutionary biology Doctoral students Olivia Graham and Morgan Eisenlord also contributed.

Co-author J. Emmett Duffy of the Smithsonian Institution was a principal investigator in a three-year, $1.3 million grant from the National Science Foundation (NSF), from which this research originated. Research and development in artificial intelligence was funded by an NSF Expeditions in Computing for Computational Sustainability grant; The initial collaboration between Harvell and Smithsonian was developed as an initiative of the Cornell Atkinson Center for Sustainability.

Gomez, who is also the director of the Institute for Computational Sustainability, and Rapazzo led the development of the Eelgrass Lesion Image Segmentation Application (EeLISA, commonly known as eel-EYE-zah), an artificial intelligence system that, when properly trained, can quickly and distinctly analyze thousands of images of seaweed. Diseased leaves of healthy tissue.

How quickly does EeLISA work? According to the researchers, it works 5,000 times faster than human experts, with similar accuracy. And by feeding the app with more information, it becomes “smarter” and gives more consistent results.

“This is really a key element,” said Rappazzo, who won the Innovative Applications Award in 2021 at the AAAI Conference on Artificial Intelligence for his work at EeLISA. “If you give the same eel assay to four different people to rank, they will all give variable measures of disease. You have all that difference, but with EeLISA, it’s not only faster but it’s named consistently.”

“In traditional machine learning, you need large amounts of data that is labeled up front,” Gomez said. “But with EeLISA, we get feedback from the scientists who provide the images, and the system improves very quickly. So in the end, it doesn’t require many classified examples.”

This project included a network of 32 field sites along the Pacific coast, spanning across 23 degrees of latitude. This diversity of regions has allowed seagrass loss disease to be studied in different climates and environments.

Thousands of images from the network of sites are fed into the EeLISA system, which analyzes each image, pixel by pixel, to determine if each image contains healthy tissue, diseased tissue or a background. EeLISA’s raw results are logged by human commentators, and corrections are made to the program so it can learn from its mistakes.

“The researchers get their results, send their corrections back to the algorithm, and it updates the next iteration,” Rapazzo said. “The original EeLISA scans to classify, when completely random, might take half an hour per scan. On the next iteration, it might take 10 minutes, then two minutes, then one minute. And we got to the point where it was human-level accuracy, It only needs to be checked intermittently.”

The AI-enabled research revealed that deviations in warm waters – regardless of the normal temperatures of a given area – were the main driver of wasting turfgrass disease. This told the researchers that studying the relationship between disease and climate change is essential for all conditions, not just seagrass meadows in warm places.

“We have invested a decade in developing disease identification tools to monitor these outbreaks on a large spatial scale,” Harvell said, because our early studies indicated that eels could be sensitive to warming-induced disease outbreaks. A crucial link in the chain of survival of fish like salmon and herring.

Gomez said the goal is to expand the scope of the EeLISA so that it can be used worldwide for “citizen science.” Aoki said that this is one of the most interesting aspects of this work.

“We can ask people to define seaweed disease in this broader way, and benefit from greater public participation,” she said. “We’re definitely a few steps away from that, but I think that’s an incredibly exciting frontier.”


Climate-induced diseases threaten the health of seaweed


more information:
Lillian R. Aoki et al, AI disease surveillance links eel wasting disease to ocean warming across latitudes, Lake science and oceanography (2022). DOI: 10.1002 / No. 12152

Introduction of
Cornell University


the quote: Artificial Intelligence Reveals Eels Vulnerability to Warming and Disease (2022, June 15) Retrieved June 15, 2022 from https://phys.org/news/2022-06-ai-reveals-scale-eelgrass-vulnerability.html

This document is subject to copyright. Notwithstanding any fair dealing for the purpose of private study or research, no part may be reproduced without written permission. The content is provided for informational purposes only.