Google Algorithm Predicts When Species Will Go 404, Not Found

(By: Wired.com)

 

Biologists have figured out the most efficient way to destroy an ecosystem — and it’s based on the Google search algorithm.

Scientists have long known that the extinction of key species in a food web can cause collapse of the entire system, but the vast number of interactions between species makes it difficult to guess which animals and plants are the most important. Now, computational biologists have adapted the Google search algorithm, called PageRank, to the problem of predicting ecological collapse, and they’ve created a startlingly accurate model.

“While several previous studies have looked at the robustness of food webs to a variety of sequences of species loss, none of them have come up with a way to identify the most devastating sequence of extinctions,” said food web biologist Jennifer Dunne of the Santa Fe Institute, who was not involved in the research. Using a modified version of PageRank, Dunne said, the researchers were able to identify which species extinctions within a food web would lead to biggest chain-reaction of species death.

“If we can find the way of removing species so that the destruction of the ecosystem is the fastest, it means we’re ranking species by their importance,” said ecologist Stefano Allesina of the University of California, Santa Barbara, who co-authored the paper published Friday in PLoS Computational Biology.

Unlike previous solutions to the coextinction problem, the Google solution takes into account not only the number of connections between species, but also their relative importance. “In PageRank, you’re an important website if important websites point to you,” Allesina said. “We took that idea and reversed it: Species are important if they support important species.”

In other words, grass is important because it’s eaten by gazelles, and gazelles are important because they’re eaten by lions.

When the researchers tested the Google algorithm against existing models for predicting ecosystem collapse, they found that the new solution outperformed the old ones in each of the 12 food webs they looked at. “In every case that we tested, the algorithm returned either the best possible solution, out of the billions of possibilities, or very close to it,” Allesina said. In this case, the “best possible solution” is the one that predicts total ecosystem collapse using the fewest number of species extinctions.

To make the circular PageRank algorithm work for food webs, which are traditionally considered unidirectional, the researchers had to solve the problem of what to do with dead ends: Not much eats a lion, but that doesn’t necessarily mean lions aren’t critical to the food chain. The scientists solved this problem by adding what Allesina calls a “root node,” which is based on the idea that all living creatures contribute to the food chain through their excrement and eventual decay.

“What we found is that the importance of a species can be connected to the amount of matter that flows to it,” Allesina said. “If species eat a lot of things, and a lot of things eat them, they tend to be important.” Previous solutions to the problem tended to underestimate the importance of species that are lower on the food chain, Allesina said, and he hopes the new solution will encourage conservation biologists to take a broader view of species extinctions.

“What I hope is that people will pick up interest and start thinking about conservation in a more network-based way,” Allesina said. “Right now, most conservationists are focused on a single species, and they just study that species. But you really have to take into account that this species is not independent, it’s really tangled in a network of multi-species interactions.”

For ecosystems on the brink of collapse, such as marine environments taxed by overfishing, Allesina said a network-based approach to conservation could make all the difference.