Predicting tipping points before they occur

— -- Say you're in a canoe on a lake, dry and firmly seated. A fish swims by and you lean over to take a look. You lean a little further, and a little further still, until, with a surge of panic you realize you're starting to fall over. You being to flail your arms, causing the canoe and you to rock wildly back and forth, but it's no good: Over the side you go.

Now you're in a new stable state — treading water next to an upside down canoe. And getting back to your former state — un-waterlogged in an upright canoe — will take a lot more work than falling out did.

You've just experienced a tipping point.

However, if you've been following work over the past 30 years in systems theory, you'll recognize this scenario as what scientists call a bifurcation point — the moment when a stable system flips over into a new stable state, after a period of rapid change.

Now some of the most prominent scientists in that field have published a new paper on detecting early warning signals before a system changes. Titled "Early-warning signals for critical transitions," the review paper is in this week's issue of the journal Nature.

Though exceedingly complex, the gist of their work is simple: detecting patterns that tend to emerge in systems just before they hit a tipping point, hopefully in time to stop the process.

"This is a very important paper," says Brian Walker, a fellow at the Stockholm Resilience Center at the University of Stockholm in Sweden.

"The big question they're trying to answer is, how the hell do you know when it's coming? Is there any way you can get an inkling of a looming threshold, something that might be a warning signal that you're getting to one of the crucial transition points?"

"The fascinating thing is that we found that very different systems react the same way and appear to obey the same universal laws as they are getting close to a tipping point," says Marten Scheffer, lead author on the paper and an ecologist at Wageningen University in the Netherlands.

Work on these theories began in the 1970s. At a meeting of researchers looking at these signals convened by Scheffer in Holland in 2007, scientists began to see examples in all sorts of places.

"We began to realize that there was really pretty cool and fundamental thing going on here," says Stephen Carpenter, one of the paper's authors and a lake ecologist at the University of Wisconsin-Madison.

Discerning these tipping points before they happen would be huge, says Carpenter.

"Managing the environment is like driving a car in the dark in the fog on the edge of a cliff. You know the edge is out there, but it's dark and foggy," he says. "We're really great at knowing where thresholds are after we fall off the cliff, but that's not very helpful."

What's fascinating about this concept of bifurcation points is that like fractals (shapes and patterns that reoccur at different sizes throughout systems and in nature), once you know about it, you realize that it's everywhere, you just hadn't realized it before.

Examples of systems that these critical transitions have been found in include:

•algae blooms in lakes

•asthma attacks

•climate change

•desertification

•epileptic seizures

•fisheries collapse

•migraines

•financial systems

In fact, research has shown that in the past eight major climate transitions in the Earth's long history, each has displayed similar behavior as it got closer to the transition point.

Ecosystems have been a special area of research using these branch of analysis.

"Sometimes you might be able to get back and sometimes you can't get back. And that's really bad, when you flip into a state when you can't get back," says Walker.

Desertification, the process by which fertile land becomes desert, is one example. Desertification often happens when a dryland area is overgrazed. At a certain point, there is too little vegetation left to maintain the soil and even if the animals are taken away, it stays in a degraded state. That's because it takes more rain to make barren patches of ground grow again than it does to maintain patchy areas of grass or brush.

The same principles seem to work in social groups. There's the famous "tipping point" where a given neighborhood becomes integrated and then when it reaches a certain percentage of one social or ethnic group, it tends to quickly "tip" to include that social or ethnic group almost entirely.

A positive example from the social sciences is the so-called poverty trap, where people are too poor to be able to invest the tiny amount of money necessary to begin trading and thus lift them out of poverty. Micro-loan programs work by shifting the system to allow that trading to begin.

Research has shown that when done properly, micro-loan programs can cause a given social group to flip from a state of extreme poverty to a better state with only a small investment of outside money.

While their work can't predict all transition points, they've begun to tease out universal principles, says Scheffer.

What the researches have shown is that these early warning signals seem to occur differently in different kinds of systems, but there's some regularity to them within a set of systems.

Systems can also being to "flicker," rapidly oscillating between to states before finally settling into a stable one.

There's also an interesting finding that as a system gets close to tipping, it becomes more like systems around it. This is seen often in financial markets and in social settings where the attitudes of individuals towards certain issues are affected by what their peers think.

Another warning signal is variability — a lot of change back and forth.

For example, in climate, "you would expect the variability of the climate to go up and down in quite an extreme way as the threshold gets near," says Carpenter. "So if the climate was nearing a change, we would expect to see a lot of records being broken. So hottest, coldest, wettest, driest. If you begin to see a pattern in those things, that tells you there's some sort of tipping point coming."

Broken feedback loops can also cause that kind of instability and "critical slowing" in a system just on the verge of collapse. One example is found in the collapse of the cod fishery off New England. The adult fish are over-harvested — but a secondary affect is that the adults were also eating other species of fish that in turn try to eat the juvenile cod.

With the adult cod gone, there's nothing left to suppress the other fish. Eventually the medium sized fish might completely wipe out the juvenile cod," says Carpenter. "Then the cod are in serious trouble because then there are no juveniles in the pipeline and the adults are being fished as well."

"There's a direct mathematical connection between crucial slowdown and increased variability it's an indicator that a bifurcation point might be coming," says Carpenter.

One problem is that finding these signals in complex systems like climate or finances requires a lot of data over a fairly long time, and that requires not only the resources to do the monitoring but also enough time before a catastrophic shift hits.

"The problem is that you may well have crossed the threshold before you've been able to pick up that change," says Walker.

While their work will be useful, "the indicators are worthless unless somebody's paying attention," says Carpenter. "You actually need to be monitoring the system in order to be able to see that a change is near."

But unfortunately financial cutbacks worldwide mean many monitoring systems are being abandoned. Carpenter cites the deterioration of the United State's Earth observation satellites.

"It's really one of the tragic ones," he says. "Fifteen years ago the U.S. had a system of Earth observing satellites that was the envy of the world. Now, we still have some, but it's deteriorating."

Finally, even if the data are there to find the patterns, managers need the capacity to respond quickly when a system is nearing a tipping point.

"They need to be empowered to act, they need the resources to act," says Carpenter. "Once you begin to see these early warning signs, the system is in a very dangerous position, you can't see the signs until it's almost too late. It's sort of an emergency room situation."