By Chris Gorski
Inside Science News Service
(ISNS) -- Development is squeezing animals into smaller pockets of land, and without sufficient planning and protection, individual animal populations could find themselves increasingly isolated.
To address this issue, researchers have been reestablishing and protecting connections on the landscape for many years, from building highway crossings to maintaining swaths of forest. These wildlife corridors are designed to enable the meanderings and migrations of animals. As scientists' efforts to improve the quality of these connections become increasingly sophisticated and more mathematical, they are finding that solving the problem has much in common with what happens when someone asks an online service to provide driving directions between two points on a map.
For planners, the goal is to preserve or create effective connections for wildlife at low cost, just as online map services aim to route travelers in the most efficient way possible. Designing a landscape to simultaneously serve the needs of multiple animal species is much more difficult because each may prefer a different type of environment. It's similar to trying to find the single best set of directions between two points for multiple modes of transportation, such as driving, walking, and mountain biking.
"Because they bring several dimensions, these problems are computationally much harder," said Carla Gomes, a computer scientist from Cornell University in Ithaca, N.Y. "If the problem is to connect just two terminals, for one species, then that problem is exactly the same computationally speaking as the problem that Google solves when I ask for the shortest path, for the fastest way to go from Boston to Ithaca, N.Y."
Michael Schwartz, a research ecologist with the U.S. Forest Service's Rocky Mountain Research Station in Missoula, Mont. had been gathering genetic data for more than 10 years and began to find that the methods they were using to analyze certain wildlife management topics were insufficient.
"We got to the point where the math became intractable to us," said Schwartz.
Schwartz started working with Claire Montgomery, a forest economist at Oregon State University in Corvallis, who had been developing methods to address both animal populations and timber management strategies.
"I was beginning to look at problems where uncertainty played a much bigger role than it had in the past in my research," said Montgomery. "And that kind of created a whole new dimension to the problem that I didn't even have a clue how to address computationally."
Multi-Purpose Land Use
Land can be managed with many different outcomes in mind. The land might be used to provide timber or to preserve native species while simultaneously being used for public recreation. Finding the best outcome for many competing interests can be complicated.
One option is providing stable habitat areas for wildlife and connecting them with corridors that enable animals to roam or migrate safely.
These competing interests make compromises inevitable. Analyzing the potential outcomes of different strategies on the inhabitants and resources that rely on a piece of land is complicated, and when the equation also includes the cost of purchasing additional land to provide those wildlife corridor areas, tradeoffs are unavoidable. Setting up a decision-making process with easily understood priorities is also important. Finding the best solution requires computational power and advanced algorithms.
"We felt pretty good about that approach for a single species," said Schwartz. "The question became, 'What happens when you look at multiple species?'"
About two years ago, Schwartz and Montgomery started working with Gomes, who is developing a new field she calls "computational sustainability." It combines aspects of ecology, economics and operations research to intensely analyze data to reveal more comprehensive solutions to difficult problems.
"You want to optimize the quality of the corridors you get for a given budget you have," said Gomes. "A lot of these problems are really highly computational."