Could our ability to predict disaster lie in maths equation?

October 23, 2006

Brussels, 20 October 2006

A European ecologist at the University of Barcelona has been studying a model to characterise the occurrence of natural disasters. Some time ago, another ecologist studying wildfires in southern Europe observed that they occurred in a logical pattern. He noticed that for approximately every three fires of any given size, there is one twice as big, and for every three fires of that size, again there is one twice as big. This phenomenon is known as the power law, and new research expands on this idea.

Despite such analysis, our understanding of the power law remains rather limited. Salvador Pueyo from the Department of Ecology at the University of Barcelona (UB) has been studying a statistical procedure to further develop our understanding of the power law. He decided to reanalyse some findings published in the journal Science concerning pests attacking coffee plantations in Mexico.

The original authors of the article studied plagues of scale insects which tap the coffee plants for sap. Scale insects exist in a symbiotic relationship with ants whereby ants protect them from predators and in turn milk them for nutritive fluid. The authors concluded that the prevalence of scale insects can evaluated using the power law, but that the power law is disrupted by the added influence of the ants. The results suggested that the power law alone is not sufficient in characterising natural phenomena.

Pueyo and his colleague Roger Jovani at the Estación Biológica de Doñana (CSIC) reassessed the data under the assumption that the power law is indeed a valid tool for analysing natural disasters. Specifically, they reworked the statistical equations of the original analysis and found that the influence of the ant species proved to be merely quantitative and not qualitative. That is to say that the number of scale insects present were affected and not the statistical distribution of their presence.

The UB ecologists established that for every ten plants with a given scale insect density there was one with twice that density where ants were not present. In the areas where ants were present they found the ratio to be 4:1.

Their new statistical model suggests that it is indeed possible to estimate ecological phenomena with "snapshot data", which they hope "might open the door to new management tools based on quantitative predictions."

In a separate article in Science last year, UB ecologists confirmed that the Mediterranean regions in Europe are the most vulnerable to disasters associated with climate change, making our ability to predict such phenomena all the more relevant.

"Current projections provide more than enough data for designing decisive action to avert disasters," says Mr Pueyo. "The Kyoto Protocol is just a minor step; surprisingly, Spain is one of the major beneficiaries of the treaty and one of the countries that is furthest away from fulfilling it.."

DG Research
Item source