Scientists now know that over 99% of the species that ever existed on Earth have already gone extinct. Dinosaurs, for example, are thought to have been wiped out around 65 million years ago when an asteroid hit our planet. However, such global extinctions remain strange: intuition tells us that isolated communities of a species should survive, but this is not the case.

Now, Ravindra Amritkar of the Physical Research Laboratory in Ahmedabad in India and Govindan Rangarajan at the Indian Institute of Science in Bangalore have tried to shed light on this puzzle. The researchers analysed existing data on populations of voles that had been preyed upon. This data was obtained in a Norwegian survey in 2000, which looked at the fate of 481 voles in different locations that had been tagged with a radio-marker. By measuring the growth rates of these populations, these results showed that the predators had a "synchronising" influence on the voles (that is, the growth rates of voles in different places began to decline in step).

Amritkar and Rangarajan then used techniques from nonlinear dynamics to simulate future behaviour and concluded that, provided there is a common forcing, separated communities of a species will synchronise together before becoming extinct. The synchronising mechanism can be likened to two grandfather clocks falling into step though subtle vibrations in the floorboards, say the researchers. These vibrations link the two clocks so they tick in phase.

Finally, the Indian scientists calculated that the likelihood of a species becoming extinct can be expressed numerically using a single parameter, which is part of a quadratic equation in the new model. The other terms in the equation refer to the population of a species, the external forcing that interacts with the populations in different locations, and the interactions between the organisms themselves. If this parameter is greater than zero, the species will die out and if it is less than zero, the species will survive. Although there are many reasons for this, one important factor is competition between animals in a species: if the animals compete, then the value of the parameter is likely to be more than zero but if the animals cooperate, it is likely to be less than zero.

Amritkar and Rangarajan say they could test for the value of this parameter in an experiment in which food is progressively taken away from a confined population of voles (or other animals, like insects). "We would then observe whether the synchronisation persists and whether the extinction is simultaneous," they say.