A computer program based on game theory could help predict the impact of human behaviour on different wildlife species.
Manchester University biological scientists Paul Higgs and Alan McKane are developing a model of evolving ecosystems called Webworld, that uses equations to describe how species compete and interact in networks of interlinked food chains.
"We want to use it to make predictions on which features of a web of species will make it more or less prone to extinction," said Dr Higgs. "It is difficult to test the model in practice, but it could be used to predict the effect humans have on ecosystems and the knock-on effect of changes we impose."
Such changes could include reducing species by over-fishing at a site, or transferring a foreign species to a new island. This would then alter the balance of existing ecosystems and force species to interact in a different way.
Wildlife charities could use the program to react early to save species threatened with extinction when their environment alters.
Webworld uses ideas from theoretical biology and statistical physics, with equations to describe the different interactions occurring between species in food webs. In the model, predators can choose which species to eat, depending on competition from other animals. Equations then decide the behaviour that is most likely to ensure survival.
The Manchester team has also used the virtual ecosystem to look at evolution. They can assess which features of an ecosystem make it most susceptible to change.
"We more or less understand that larger-scale extinction events come from external causes like meteor strikes killing off dinosaurs. Many scientists believe evolving systems build up towards an unstable point beyond which a huge change occurs causing mass extinction," Dr Higgs said.
But the model has disproved this theory, showing instead a continual turnover of species as moderate extinction events take place. The large avalanches of extinction known as "self-organised criticality" were not seen.