X Close

GEE Research

Home

Research in Genetics, Evolution and Environment

Menu

Archive for the 'Award' Category

Award-Winning Bat Conservation

By Claire Asher, on 16 September 2013

This year’s Vincent Weir Scientific award for bat conservation biology has been awarded to GEE’s Charlotte Walters for her PhD work on the iBatsID tool.

The Vincent Weir Scientific Award is an annual award given to a UK-based student for their outstanding contribution to the conservation biology of Bats. It is awarded by the Bat Conservation Trust (BCT), a national organisation devoted to the conservation of bats and their habitats within the UK. Charlotte Walters, who recently completed her PhD with the Zoological Society of London (ZSL), University College London (UCL), University of Kent and BCT, has been awarded the prize for her contribution to bat conservation and particularly her work for the Indicator Bats Program (iBats).

iBats is a partnership between ZSL and BCT, aiming to monitor global changes in bat biodiversity and provide valuable data for policy makers and conservation groups. They provide training and equipment to projects monitoring bat biodiversity to ensure standardised methodology which will enable global comparisons. They have also developed a number of free tools for iPhone and Android which enable fast, simple and efficient detection and identification of bats, and Charlotte’s iBatsID program is a key part of this.

Myotis bechsteini
Image Credit: Gilles San Martin, used under creative commons licence.

During her PhD, Charlotte developed the iBatsID tool, an automatic tool for acoustic identification of European bat ecolocation calls. The tool is able to identify 34 different species of bat based on their calls alone, and is enabling scientists to achieve consistent monitoring of bat populations across Europe. The tool uses ensembles of artificial neural networks to classify bat echolocation calls and identify which species or group the call belongs to. Dr Karen Haysom (Director of Science, BCT) says “New tools and techniques to assist monitoring help us find out more about these fascinating and vulnerable creatures, [and] Charlotte particularly impressed the judges with the innovation and technical quality of her research”.

Eptesicus nilssonii

Bats are ecologically important, playing a key role as predators and seed dispersers. They are also very sensitive to human activities, and are useful as ‘indicator species’ for monitoring biodiversity patterns in general. In Europe, all 52 species of Bat are protected by law as part of the “Agreement on the Conservation of Populations of European Bats“. However, being nocturnal and generally small, they are difficult to detect visually or by trapping. Recording bat calls can allow researchers to survey difficult habitats and gain a clearer picture of what bat species are present and in what numbers. But a standardised statistical method for identifying the species of bat based upon it’s call was needed. This has previously been difficult to achieve, but the recent publication of a global library of bat calls, EchoBank, enabled this type of large-scale identification project to be attempted.

Bat calls vary between species and have been shaped by natural selection relating to species’ ecology. However, calls also vary between individuals within a species according to sex, age, habitat and geographical location, and social environment. Bats also vary their calls depending on what they’re doing – calls are longer when a bat is searching for prey and become shorter as it narrows in on it’s target. So, identifying a species by it’s call is a little more complex than one might expect. Charlotte developed an artificial neural network which was trained on calls of known species and can then be used to identify new calls recorded in the field.

Example of an Artificial Neural Network
Image by Chrislb, used under creative commons licence.

Artificial neural networks are computer models inspired by the central nervous system of animals. They are represented as an interconnected set of ‘neurons’, each of which makes simple calculations which together generate complex behaviour. Artificial neural networks are ‘trained’ first and this training determines the simple algorithms performed by each neuron. The trained network can then be used on real data. In the case of iBats, this involves training the network using calls for which the bat species is known, and the finished neural network can then be used to estimate which species an unknown recorded call belongs to. ANNs are a form of computer learning, and will improve in their accuracy with training – the network of neurons is able to ‘learn’ from it’s mistakes and refine the algorithm to improve classification. This method proved to be highly accurate; 98% of calls from 34 species can be accurately classified into a ‘call-type’ group, and 84% can be classified to species-level.

The iBatsID tool is freely available online, enabling researchers to utilise a standardised methodology for identifying bat species across Europe. This will facilitate large-scale comparative studies and will be particularly useful for studying European bats that have a large geographical range or are migratory. This data will be important for making conservation decisions for the future, and is therefore crucial for bat conservation but also for biodiversity monitoring in general, as bats can provide an accurate assessment of the health of entire biological communities.

Original Article:

() Journal of Applied Ecology

logo-bct

This research was made possible by funding from the Natural Environment Research Council (NERC) and the Bat Conservation Trust