Newswise — Algorithms can predict what movies or songs you might like, but they can also predict which species a predator would most likely eat.
Scientists at Flinders University’s Global Ecology Lab employ machine learning to detect species interactions and forecast the probable extinction of species, enabling proactive planning of intervention before it occurs.
"The Earth confronts an ecological emergency, as climate change, invasive species, habitat degradation, and various anthropogenic actions trigger numerous extinctions," remarks Dr. John Llewelyn, a Research Fellow at Flinders University’s College of Science and Engineering.
"Numerous extinctions arise from species interactions, influenced by the gain or loss of interactions with other species. We have discovered that machine learning can anticipate the predator-prey dynamics within an interconnected ecosystem," explains Dr. John Llewelyn, a Research Fellow at Flinders University’s College of Science and Engineering.
Dr. Llewelyn clarifies that 'co-extinctions' are instances of extinctions triggered by the decline or disappearance of other interconnected species. For example, when a predator loses its prey and subsequently goes extinct.
In contrast, invasive predators like cats, foxes, and brown tree snakes have the potential to induce extinctions in native prey species that lack prior experience dealing with such predators.
"Such extinctions occur when vulnerable species come into contact with novel predators. Therefore, understanding the interactions between species is crucial for predicting and preventing future extinctions," emphasizes Dr. Llewelyn. "Yet, currently, our knowledge is limited to a minuscule portion of the existing or potential interactions, especially concerning invasive species. This limitation poses challenges in accurately predicting extinctions."
The recent research conducted by the Flinders University team has revealed that machine learning methods can effectively utilize species' characteristics to accurately predict predator-prey interactions in birds and mammals. By identifying interacting species, machine learning offers the potential to forecast and, ideally, prevent extinctions before they occur.
The algorithm learns the correlation between species traits and their interactions by analyzing data on species that do interact, species that do not interact, and the corresponding traits of those species. This form of artificial intelligence can subsequently be supplied with a list of species and their traits to predict the interactions among the species in the given list.
“We can use this method to fill in the many gaps we have in our knowledge of species interactions,” says Dr Llewelyn.
These gaps encompass undocumented interactions that are currently occurring, interactions between species that have long been extinct, and interactions that invasive species would have if introduced to a new geographic area.
"Understanding the interactions between species enables us to comprehend how environmental disruptions, such as climate change and the introduction of non-native species, can trigger cascading effects within ecological communities. This knowledge empowers us to gain insights into the mechanisms behind extinctions," explains Dr. Llewelyn.
Species interactions are of utmost importance in ecological systems; however, comprehensive data on these interactions are scarce across most ecological communities. This limitation hampers our ability to accurately predict how ecosystems operate and respond to disturbances.
Dr. Llewelyn emphasizes that human beings are entirely reliant on biodiversity and thriving ecosystems. Therefore, we have a profound responsibility to preserve biodiversity not only for its intrinsic value but also for the numerous benefits it bestows upon human societies.
RSS