Ecologists are taking a page, and its ranking, from Google.
A new algorithm inspired by the search engine works well for predicting which species losses will trigger the fastest collapse of a food web, says theoretical ecologist Stefano Allesina of the University of Chicago.
Food webs describe the pattern of what eats what in the neighborhood. If one kind of grass or bug, for example, disappears, creatures that fed on it would need to find something else for lunch. If they couldn’t, or if the alternative entrées went extinct too, then the loss could trigger a cascade of extinctions. Losing certain species can starve so many others that the whole food web unravels.
The new algorithm works much like PageRankTM, Google’s system for ranking the importance of Web pages. The food web version did a better job of predicting collapse than simply comparing the number of connections each species has with other species in the food web. The method also beat out analyzing the network to find hub species, Allesina and Mercedes Pascual of the University of Michigan in Ann Arbor report online September 3 in PLoS Computational Biology.
“The problem of how ecosystems are likely to respond to the loss of species is quite important, particularly in light of how many different ways human activities are resulting in the local extinctions of populations,” says computational ecologist Jennifer Dunne of the Santa Fe Institute in New Mexico. Global warming, the introduction of species such as the zebra mussel, development that destroys habitat, pollution and plenty of other menaces make food web vulnerability an urgent concern, she says.
Allesina got the idea for treating food webs like the World Wide Web while he was at the National Center for Ecological Analysis and Synthesis in Santa Barbara, Calif., and chanced upon a description of Google’s page ranking system. “I said, ‘That looks familiar,’’’ he remembers. In essence, the system calculates a page’s importance, or value to searchers, depending on the importance of the pages that link to it. Through the magic of mathematics, it works. In a food web, species draw importance from the importance of the species that eat them.
The Google ranking system, though, essentially assumes that any page might lead to any other page. But that doesn’t make sense for food webs. “Energy does not go from the grass to the lion without going through the zebra,” Allesina says. He and Pascual only made links between predator and prey. And for every species the researchers added a path to a “detritus pool.” All species can thus die, decay and become nutrients for plants or the other primary producers of food in the web.
To compare their algorithm with others, the researchers used information from real-world food webs. One of the other programs, called a genetic algorithm, provided a gold standard. The new algorithm matched the genetic algorithm’s results without its heavy computational demands, Allesina and Pascual report.
Dunne calls the approach a “novel, exciting contribution.” Now, she says, it would be interesting to try to bring the algorithm closer to real life by adding factors such as the relative amounts of energy flowing through the connections.