THEY were once used down coalmines to warn workers if there was poisonous gas around – if they died then the miners knew it wasn’t safe.

But now, after saving countless miners’ lives, the humble canary is being hailed as a potential saviour of entire ecosystems up to three decades in advance – by simply leaving the area.

A new study shows that canaries are often the first species to disappear from an area which is showing signs of a catastrophic ecosystem tipping point – up to 30 years in advance. Scientists discovered after focussing on three lakes in China, that the canary was always one of the first to disappear, which can now be used as a potential warning sign for the future.

The study, published in the journal Ecology, looked at canary numbers in three lake ecosystems in China.

The research showed changes in the abundance of species from algae (diatoms) and aquatic midges (chironomids) communities as they compete for resources under environmental pressures.

From this data it was possible to identify three types of organism including slowly replicating but strongly competitive ‘keystone’ species, weakly competitive but fast replicating ‘weedy’ species and slowly replicating and weakly competitive ‘canary’ species.

It found as environmental degradation impacts on the ecosystem, keystones initially prevail through competitive dominance over others, resulting in the early demise of canaries. With continuing degradation affecting all species, this leads to the eventual collapse of the keystone species as they are replaced by the weedy species. The loss of keystones puts the ecosystem into a critical transition – the point at which a system tips into an alternate state from which it can be very hard to recover.

Lead author Professor Patrick Doncaster said: “We identified an early warning signal of these changes in the relationship between compositional disorder, which is roughly equivalent to how jumbled up a community of species is, and biodiversity, which measured the total amount of species in a community.

“How these populations change over time reveals statistically significant signals of an approaching tipping point. In some cases we could detect an early warning signal up to three decades before the actual tipping point.”