By Emma Marris

Could you prevent nine local extinctions by hastening one extinction? It sounds completely counterintuitive, but a pair of ecosystem modelers are proposing that conservationists could sometimes prop up a troubled ecosystem by removing one or more of its species--and using models to determine the timing and order of those removals.

The species that make up an ecosystem are connected in complex "food webs" of eater and eaten. When one species disappears, its predators can no longer eat it and its prey are no longer eaten by it. Changes in these populations affect others. Such impact 'cascades' can be unpredictable and sometimes catastrophic.

Sagar Sahasrabudhe and Adilson Motter of Northwestern University in Evanston, Illinois, have shown that in several model food webs, as well as in two webs modeled with data derived from real ecosystems--the Chesapeake Bay off Maryland and Virginia and the Coachella Valley in Southern California--removing or partially suppressing one or more species at key time points after one member has gone extinct saves other members of the web from local extinction. They publish the results January 25 in Nature Communications.

The idea relies on the fact that ecosystem networks can often shift to a different stable arrangement after losing members. "Ecological systems are quite robust, actually," says Motter. The famous "balance of nature" is perhaps better understood as the "multiple possible balances of nature." But the order of removals matters.

Remove A and then B, for example, and a given web might change shape but retain all its other members; remove B and then A, however, and the cascade of changes drives many of the other members extinct.

Exit strategy

In very simple webs, the impacts can be easy to follow. For example, the removal of a large predator could allow a medium-sized predator to increase in numbers and eat its smaller mammal and bird prey into extinction. In this case, keeping a lid on the numbers of the medium-sized predator would prevent these extinctions.

But even small webs can harbor complexities that can make the order of removals for ecosystem stability challenging to sort out. Motter likes the story of the island foxes (Urocyon littoralis) on the Channel Islands off the coast of California, which is recorded in the study. When feral pigs were introduced to the islands, they attracted golden eagles that preyed on both pigs and foxes. Fox numbers then dropped. Removing the pigs would have left the foxes as the sole diet of the eagles, and likely doomed them. So instead, conservationists captured and relocated the eagles and only then eradicated the pigs. The fox population is now recovering.

"The same actions at different times have very different consequences," says Motter.

In more complex webs, the key species that needs to be removed or suppressed to head off more serious collapse isn't intuitively clear. But, by modeling a variety of food webs using established ecological principles, the researchers were able to find such species and they hope that the algorithms that they created might eventually be able to identify target species in the real world.

Factoring in complexity

However, the results will only be accurate if the real ecosystem is well represented by the model. In their food-web modeling, Sahasrabudhe and Motter have used accepted ecological models of predator-prey relationships, but a more elaborate representation of an ecosystem would also include parasitism, seed dispersal, competition, mutualisms (in which species make life easier for each other), nutrient dynamics and more. And to include such complex detail in a model, scientists will first have to go out and gather that information in the field. Who is eating whom? Who is pollinating whom?

When the algorithms point to an exotic species as a target for removal or suppression, conservationists are likely to have little problem with the idea. But if a native species is the proposed target, that will go against many conservationists' impulses to protect rather than remove.

Neo Martinez, director of the non-profit Pacific Ecoinformatics and Computational Ecology Lab in Berkeley, California, says that Sahasrabudhe and Motter's ideas are exciting, but the conservatism of conservation means that they won't be relied on in isolation, at least not right away. "Because of the lack of realism -- we don't include everything in these models -- no one is going to make an important conservation decision solely on these models. That is a long time in the future."

But a long time in the future isn't never. Martinez says that whereas six or eight years ago ecologists generally considered ecosystems too complex to ever be productively modelled, not unlike the stock market, today modellers are gaining confidence.

Motter agrees. "In the long run, I think we will have people in the field advocating for the suppression of native species." He points out that land managers are already doing so, less systematically, by running regulated hunting of prey species in areas where top predators have been extirpated. Human impacts are just too great on most ecosystems, he says, for us to just hope they will sort themselves out. "In the presence of perturbations, it is reasonable to consider compensatory perturbations," he says.