Bat Identification.

Most of my readers know that I have an interest in bats. Bat recognition is a difficult task. A bat detector helps to give some idea of the bat species. However you also need to see the bat as well as monitor the frequency of its calls. You can observe the flight pattern as well as the size of bats to help with identification. When I go out on a night, I also take a 1 million candlepower torch. When I detect a bat, I can also at the same time illuminate it. So I can see the size, markings and flight pattern. Light does not bother the bats. They often hand around bright light sources such as street lamps to pick of insects like moths that are attracted to the lights.

All this is about to change. New research has made identification much easier from their ultra-sound alone.

Just as differences in song can be used to distinguish one bird species from another, the pips and squeaks bats use to find prey can be used to identify different species of bat. Now, for the first time, a team of ecologists, including Professor Gareth Jones from the University of Bristol, have developed a Europe-wide tool capable of identifying bats from their echolocation calls.

The new free online tool – iBatsID – will be a major boost to conserving bats, whose numbers have declined significantly across Europe over the past 50 years. Details are published today in the British Ecological Society's Journal of Applied Ecology.

Scientists from the Zoological Society of London, University College London, University of Auckland and 9 other institutions worked together to amass echolocation calls from 34 different species from all over Europe, forming part of EchoBank, a global echolocation library of more than 200,000 bat calls.

The calls were then analysed to find out which characteristics were most useful in distinguishing different bat species. According to lead author and PhD student Charlotte Walters: “Lots of different measurements can be taken from an echolocation call, such as its maximum and minimum frequency, how quickly the frequency changes during the call, and how long the call lasts, but we didn’t know which of these measurements are most useful for telling different species' calls apart.”

The 12 most useful call parameters were then used to train artificial neural networks to produce the new identification tool, iBatsID, which can identify 34 different bat species across the whole of Europe. Most species can be identified correctly more than 80 per cent of the time, although accuracy varies because some species are much harder to identify than others.

Later.....