Research published this week in Science offers the clearest picture yet of the resounding consequences of the decline of land mammals on food webs over the past 130,000 years.
It’s not a pretty picture.
“While about 6% of terrestrial mammals disappeared during this time, we estimate that more than 50% of mammalian food web links disappeared,” said the ecologist. Evan Frickmain author of the study. “And which mammals are most likely to decline, both in the past and now, is key to the complexity of the mammalian food web.”
A internet food contains all the links between predators and their prey in a geographical area. Complex food webs are important for regulating populations in ways that allow more species to co-exist, supporting biodiversity and ecosystem stability. But animal decline can degrade this complexity, undermining the resilience of ecosystems.
Although mammalian decline is a well-documented feature of the biodiversity crisis – with many mammals now extinct or persisting in a small part of their historical geographic ranges – it is unclear how much these losses have degraded food webs. global.
To understand what has been lost in the food webs connecting land mammals, Fricke led a team of scientists from the US, Denmark, UK and Spain using the latest machine learning techniques to determine “who ate who” 130,000 years ago. Until today. Fricke conducted the research while on a fellowship at rice university and is currently a researcher at the Massachusetts Institute of Technology.
Using data from modern observations of predator-prey interactions, Fricke and his colleagues formed their machine learning algorithm to recognize how species traits influenced the likelihood of one species feeding on another. Once trained, the model could predict predator-prey interactions between species pairs that have not been directly observed.
“This approach can tell us who’s eating who today with 90% accuracy,” said Rice’s ecologist. Lydia Beaudrot, the lead author of the study. “It’s better than previous approaches have been able to do, and it’s allowed us to model predator-prey interactions for extinct species.”
The research provides an unprecedented global view of the food web that connected Ice Age mammals, Fricke said, as well as what food webs would look like today if saber-toothed cats, ground sloths giants, marsupial lions and woolly rhinos still roamed alongside the surviving mammals. .
“While fossils can tell us where and when certain species lived, this modeling gives us a richer picture of how those species interacted with each other,” Beaudrot said.
By mapping the evolution of food webs over time, the analysis revealed that food webs around the world are collapsing due to animal decline.
“Modeling showed that terrestrial mammal food webs have degraded much more than you would expect if random species had gone extinct,” Fricke said. “Rather than resilience under extinction pressure, these results show slow-motion food web collapse caused by the selective loss of species central to the food web.”
The study also showed that all is not lost. While extinctions have caused about half of reported food web declines, the rest stems from shrinking geographic ranges of extant species.
“Restoring these species to their historic range offers great potential to reverse these declines,” Fricke said.
He said efforts to recover native predators or prey, such as the reintroduction of Lynx in Colorado, European bison in Romania and fishermen in Washington State, are important in restoring food web complexity.
“When an animal disappears from an ecosystem, its loss ripples through the web of connections that connect all species in that ecosystem,” Fricke said. “Our work presents new tools for measuring what has been lost, what more we stand to lose if threatened species go extinct, and what ecological complexity we can restore through species recovery.”
Study co-authors include Chia Hsieh and Daniel Gorczynski rice, Owen Middleton from the University of Sussex, Caroline Capello from the University of Washington, Oscar Sanissidro from the University of Alcalá, John Rowan from the University of Albany and Jens-Christian Svenning from Aarhus University.
The research was funded by Rice University, the Villum Foundation (16549) and the Independent Research Fund of Denmark (0135-00225B).