Newswise — Algorithms can predict which movies and songs you might like, but they can also predict which species predators are most likely to eat.
Researchers at Flinders University’s Institute of Geoecology can use machine learning to identify species interactions and predict which species are most likely to go extinct, so they can intervene before extinctions occur. can now be planned.
“The planet faces an environmental crisis with mass extinctions caused by climate change, invasive species, habitat loss and other human-related activities,” said Dr. John Llewellyn, a research fellow in the Flinders School of Science and Engineering. .
“Many of these extinctions are mediated by species interactions, caused by the loss or gain of interactions with other species. And machine learning is predicting who eats who in a connected world of species.” We have found that we can predict the
Co-extinction, Llewellyn says, is extinction caused by the decline or extinction of other interacting species, such as predators that go extinct after losing their prey.
Conversely, invasive predators such as cats, foxes, and pythons can exterminate immature native prey that have never dealt with similar predators in the past.
“These extinctions are caused by vulnerable species interacting with emerging predators, and knowing which species interact is essential to predicting and avoiding future extinctions. added Dr. Llewellyn. “However, at present, we know only a fraction of the species interactions that occur, or in the case of exotic species, the species interactions that may occur, making extinction predictions difficult. “
New research from the Flinders team finds that machine learning techniques can use species traits to accurately predict predator-prey interactions in birds and mammals. By identifying interacting species, machine learning can help predict and hopefully prevent extinction in advance.
From information about which species interact, which do not, and the traits of the species involved, the algorithm learns how traits relate to species interactions. If you give this type of AI a list of species and traits, it can predict which species in the new list will interact.
“Using this method, we can fill many gaps in our knowledge of species interactions,” says Dr. Llewellyn.
These gaps include undocumented interactions occurring today, interactions between long-extinct species, and interactions that would occur if alien species were introduced to new regions. .
“Knowing which species are interacting allows us to identify how environmental disturbances such as climate change and introduced species have cascading effects on ecosystem communities, and how extinctions occur. You will understand.”
Species interactions play a fundamental role in ecosystems, but few ecosystem communities have complete data describing such interactions and how ecosystems function and are perturbed. obstacles in predicting how they will respond to
“Humans are totally dependent on biodiversity and healthy ecosystems, so we have a responsibility to maintain biodiversity not only for the benefits it brings to human societies, but also for itself.” Dr. Llewellyn says.
