Thursday, 3 May 2018

Twilight Bat Walks

by David Jackson from Mid Essex Hospital Services NHS Trust

In the United Kingdom we have 18 species of bat, all are insectivorous and a great environmental indicator. A single bat can eat up to 3000 midges in one night making them an excellent natural insect controller, but unfortunately over the last century their populations have declined, making each roost important for future survival. Due to their decline, bats and their roosts are protected under UK and European law with a roost defined as any place that a wild bat uses and is protected whether bats are present or not.

At Mid-Essex Hospital Services NHS Trust, we are proud to have high bat activity around our Broomfield Hospital site, with a bat roost of Soprano Pipistrelles known and monitored for several years.  Both Soprano & Common Pipistrelles have been found foraging within our natural areas, feeding on insects during their summer months and highlighting our effective natural management on site. 

Bat walks offer a unique educational opportunity to experience these intriguing creatures, and in 2016 we launched an inaugural Twilight Bat Walk Programme, led by a local bat enthusiast.  The walks proved to be a huge success, providing education in an endangered species whilst promoting healthy walking alternatives within Broomfield Hospital and community.  Following on from these walks, a larger Twilight Programme was produced for the summer of 2017.  

A stunning 400% increase in participation occurred, with over 100 individuals expressing interest, comprising of staff, patients and members of our local community.  David Jackson, the bat enthusiast, and now the Trust’s Sustainability Project Coordinator said ‘It’s been great to see the walk participation increase so quickly…educating those in our community about endangered species is hugely valuable to their future conservation, whilst achieving it through the promotions of an active lifestyle.’ 

Supported by the Essex Bat Group and Heart & Sole Walking Scheme the 2018 Twilight Programme is now available.  Join us with friends and family to experience the hospital site bat activity and roost.  What will you find? 

The Bat Walk Dates 
25th May, starts 8:30pm 
13th July, starts 8:45pm 
27th July, starts 8:25pm 
17th August, starts 7:45pm 
7th September, starts 7:00pm

Spaces are limited, on a first come first serve basis.  If you’d like to express an interest please contact David Jackson on or 01245 514558.  

Friday, 23 March 2018

Bats in Greywell tunnel, Hampshire

by Paul Hope

Greywell tunnel is a disused canal tunnel on the Basingstoke Canal in north Hampshire. The 1,124m long brick lined tunnel was constructed in tandem with the canal between 1788 and 1794. The tunnel is approximately 5m high and has a maximum width of 4.5m. It was excavated through an area of changing geology from Hampshire chalk through Reading beds into London clay. There was a partial collapse in the tunnel during the 1930s and a further fall during the 1950s that left the bore of the tunnel blocked about 800m from its eastern entrance and at 130 m from its western entrance. Bat friendly grilles prevent public access to the site.

Greywell tunnel has long been known as an important site for hibernating and swarming bats. It was designated a SSSI for bats in 1985. Hibernation surveys are conducted from a boat by a team of up to five surveyors. As you work your way into the tunnel light from the portal becomes a distant dot. During hibernation counts many of the bats encountered are hanging against the brick lining, however a high proportion are often packed into crevices and openings in the brickwork left during the tunnels construction. Within crevices it’s sometimes a matter of counting ears and feet to get as accurate a count as possible and on occasion it’s not possible to record bats to species.
Natterer's bats at Greywell. (c) Roy Champion

Natterer’s bat is the most commonly encountered species (highest count of 569 in 2013), Daubenton’s bat is also regularly recorded (highest count of 69 in 1986). We also often record whiskered, Brandt’s and brown long-eared bats. The count this January gave some very exciting results. We had our second highest count for Natterer’s bat in the eastern end (564) and our first ever record of a greater horseshoe bat. The horseshoe bat is an excellent record for North Hampshire. Bat Group records show it is the seventh on record; this includes two records for Winchester from the 1940s.
Greater horseshoe bat at Greywell. (c) Roy Champion

Sadly after over 30 years of monitoring at the site, future regular monitoring is threatened by a recent move by the owners Hampshire County Council to declare it a Confined Space. Now surveys can only be conducted once every two years due to the excessive costs of gas monitoring equipment and a rescue team stationed outside the tunnel. We do not feel that this is often enough to provide adequate data resolution for the SSSI condition assessment, but Natural England have said that it will be sufficient.

Thursday, 15 March 2018

Using acoustics to assess biodiversity in cities

Cities now support over half of the world’s human population. They also support endemic and threatened wildlife. This wildlife provides multiple ecological services to human urban populations, and also provides the daily contact with nature that most of us now experience. However, our understanding of how wildlife persists in urban environments, and how we can design and manage cities to provide the best possible habitat for wildlife is still not very well understood. One reason for this is due to the difficulties of assessing biodiversity, especially in cities. Human surveyors have to consider safety issues, survey equipment is vulnerable to vandalism, theft and destruction, and getting access to private land to conduct surveys can be very bureaucratic. Through my Engineering Doctoral research (EngD) I developed new acoustic tools that can be used to make it easier to assess biodiversity in urban environments. These tools should hopefully be used in the future to better understand the biodiversity supported by cities and to inform the design and management of cities for biodiversity.

Acoustics and biodiversity assessment

Lots of wildlife produce sounds, either by vocalising – think a singing songbird - or through movements of their bodies – think the hum from the rapid wingbeats of a bumblebee. Some species also make sounds that cannot be heard by humans, such as the ultrasonic echolocations of bats.
Ornithologists have used the species-specific calls emitted by birds to identify calling species for centuries. More recently, ultrasonic recorders have been used to record the species-specific echolocation calls of bats which can be used to identify species.
But these examples are very species-specific. There are other species that live in cities such as insects and non-flying land animals like foxes and squirrels that also make sound.
The field of ecoacoustics uses the sounds emitted by all species to get a measure of whole ecological communities, rather than focussing on specific species and species groups. Ecoacoustic theory posits that the soundscape (sounds at a landscape scale) of an environment is composed of biophony (sounds emitted by non-human organisms), anthrophony (sounds associated with human activities) and geophony (abiotic sounds such as wind and rain).
The three elements that compose a soundscape: biophony, anthrophony and geophony. Image credit: Pijanowski et al. (2011) BioScience.

New technology for ecoacoustics

The use of ecoacoustics to monitor biodiversity has really been made possible by the development of passive acoustic recording technology, such as the very popular products from Wildlife Acoustics. These are weather-proof acoustic recorders that can be left outside for long periods of time, recording sound on a user-defined schedule. This technology is making it possible for biodiversity scientists and conservationists to collect vast amounts of acoustic recordings.

An SM2+ SongMeter is an example of a passive acoustic recorder that is commonly used in ecoacoustic research. These devices can be deployed in the environment recording at a pre-defined schedule for days/weeks/months at a time. Image credit:

But what to do with all this acoustic data? It is not practical or efficient to spend days/weeks/years listening to the recordings that have been made.
To overcome this Big Data problem, ecoacousticians have developed algorithms that can measure the biotic sound in acoustic recordings producing a proxy measure of biodiversity. These algorithms are called acoustic indices and can be used to quickly produce measures of the biotic sound in large volumes of acoustic data. They measure a few characteristics of the acoustic data, such as the amount of sound at particular frequencies, to produce a summary measure of biotic sound within an entire sound recording.

Acoustic indices produce measures of the biotic sound in audio recordings. Four commonly used acoustic indices include: A) Acoustic Complexity Index (ACI, Pieretti et al. 2011 Ecol. Indic.), B) Acoustic Diversity Index (ADI, Villanueva-Rivera et al. 2011 Landscape Ecol.), C) Bioacoustic Index (BI, Boelman et al. 2007 Ecol. Appl.) and D) Normalised Difference Soundscape Index (NDSI, Kasten et al. 2012 Ecol. Inform.). The ACI sums the absolute difference in signal power within frequency bins over time using a sliding window and defined temporal steps (indicated by arrow). The ADI is calculated as the Shannon’s diversity index for each recording based on the signal power occupancy of each 1 kHz frequency band. The BI calculates the signal power within 2-8 kHz frequency band of recordings. The NDSI calculates the ratio of signal power in the frequency bands between 1-2 kHz and 2-8 kHz to measure the level of anthropogenic disturbance on the landscape.

Ecoacoustics in the city

However, ecoacoustic research and the development of acoustic indices has tended to focus on less disturbed environments than cities, such as temperate woodlands, coastal forests and Mediterranean scrub. The challenge of my EngD research was to see if ecoacoustic could be applied in this new, highly anthropogenically disturbed environment.
I started off by assessing the suitability of a suite of commonly used acoustic indices for use in the urban environment. I collected low (0-12kHz) and high (12-96kHz) frequency audio recordings from 15 churchyard sites across the Greater London area. I was lucky to collaborate with the Diocese of London on this data collection while they conducted a London-wide ecological survey of their churchyards.

Setting up my acoustic sensors

To test exactly what sounds the acoustic indices measured in the acoustic data, I listened to and manually labelled the sounds in a random selection of my recordings. To do this, I co-developed AudioTagger, a bespoke audio analysis software that allows you to quickly listen to and view audio recordings, and draw labelled bounding boxes around sounds on spectrograms (a visual representation of an audio recording).

AudioTagger in action. Sound recordings can be quickly listened to, viewed as spectrograms and annotated by drawing labelled bounding boxes around sounds of interest. Sounds labelled here include birds (blue boxes), electrical buzzes (pink) and road traffic (red).

I identified a very wide range of sounds in my recordings. Anthropogenic sounds dominated the dataset and the variety was much greater than in previous ecoacoustic studies, including sounds such as road traffic, sirens, church bells, footsteps, and applause. Biotic sounds were mainly made by birds and bats, and geophonic sounds were either wind or rain.
Of the four acoustic indices I tested, either the measures of the acoustic indices did not correlate with the amount of biotic sound in recordings, or were biased by non-biotic sounds in recordings (Fairbrass et al.2017).

Therefore, I would not recommend that any of the acoustic indices I tested are used to assess biodiversity in urban environments.

Machine learning as an alternative to acoustic indices

Machine learning algorithms learn to recognise patterns in data based on examples that they’ve seen previously, for example how a spam email application learns to filter emails based on what a user has previously marked as spam (skip to 32:38 for a great introduction to machine learning).
Deep learning algorithms (a type of machine learning) choose by themselves what characteristics define different groups of data, rather than relying on humans to choose. Therefore they can use many more parameters within data to characterise groups than a human ever could, making them extremely powerful.
Machine learning algorithms are potentially much more powerful than acoustic indices, as they do not rely on human defined characteristics of data. What they do rely on is having a large dataset of labelled data with which to learn the characteristics of different groups.
To train a deep learning algorithm that could measure biotic sound in audio recordings from the urban environment without being biased by the non-biotic sounds in the city, I collected audio recordings from 63 sites across the Greater London area.
I used AudioTagger to label all the biotic, anthropogenic and geophonic sounds in a random selection of 45 minutes of recordings from each of the 63 sites. This labelled data was used to train and test a pair of Convolutional Neural Network algorithms, CityNet (Fairbrass Firman et al. in review), which produce a measure of biotic and anthropogenic acoustic activity in noisy acoustic recordings from the urban environment.

Audio data recorded at 63 green infrastructure sites in and around Greater London (A) was used to train (black dots) and test (red dots) the CityNet system. CityNet can be used to summarise the biotic and anthropogenic sounds in large volumes of audio data from noisy urban environments. A week of audio recordings are summarised here to produce daily average patterns of biotic and anthropogenic sound (B and C), which can be interesting to compare between different types of environments, such as sites in the centre (B) and on the outskirts (C) of the city. Image Credit: M. Firman

The algorithms can be used to produce summaries of the biotic and anthropogenic sound in large volumes of audio data from cities – at you can explore the patterns of biotic and anthropogenic sound at my 63 study sites.

Biotic sound as a proxy measure for biodiversity

An open question that remains in ecoacoustics is whether biotic sound is a good proxy measure for biodiversity? And does more biotic sound mean more or better biodiversity? Some have argued that due to the way species avoid calling at the same time and frequency, more sound does correlate with more diversity, but that this relationship only holds in the tropics (Krause & Farina 2016).
To investigate whether biotic sound can be used as a proxy measure of habitats in cities, I conducted habitat surveys at all my study sites and compared local and landscape habitat measures with the biotic and anthropogenic acoustic activity and diversity.
Initial results suggest that there is a relationship between the biotic sound recorded at a site, and the characteristics of the local and landscape habitat. However, this research needs a bit more work before I’m confident in the results, and I will continue to work on this to try and answer the question of whether biotic sound can be used as a proxy measure of habitats in cities.

Next steps

In this research I have found that existing ecoacoustic tools in the form of acoustic indices are not suitable for use in cities as they either do not measure biotic sound or are biased by non-biotic sounds in recordings (Fairbrasset al. 2017). To overcome the shortcomings of acoustic indices, I have developed deep learning tools, CityNet (Fairbrass Firman et al. in review), which measure the biotic and anthropogenic sound in acoustic recordings made in noisy urban environments. Finally, I have found that biotic sound may be a good proxy for the characteristics of habitats in cities.
Cities are exciting places in which to develop new technologies for assessing biodiversity because the availability of power and Wifi connectivity makes it possible to develop autonomous systems for monitoring wildlife. For example, smart sensors and machine learning algorithms have been used to develop the world’s first end-to-end system for monitoring bat populations in the urban environment ( ). I think it is important that machine learning and sensing technologies continue to be used to improve our understanding of wildlife in cities, enabling us to design and manage better cities for the future.

Monday, 12 February 2018

Daubenton's on ice!

From BCT's Scottish Officer, Liz Ferrell, comes a fascinating story of winter bat behaviour:

Graham White, a nature enthusiast, had never seen a Daubenton’s bat before and certainly not during daylight hours! He therefore felt very lucky to see one (on two separate occasions) over the festive period where he lives in Coldstream, the Scottish Borders. Graham told us he had been walking along a section of the River Tweed on the 4th December. It was south facing and the bank was getting a lot of warmth - even on a very cold day. The old mill weir also meant the river was slow flowing and about a metre deep. The result was like a mirror with not even a ripple on the water. The only movement came from insects which dotted the surface. And then there it was, a Daubenton’s bat flying over the water at one o’clock in the afternoon!
By pure chance, on the 7th January, Graham saw another Daubenton’s. This time 1.5 miles from the River Tweed flying over Hirsel Lake along with the geese, swans and goldeneye he had been watching. The bat flew right next to Graham and for a further two minutes he watched and photographed the bat before it disappeared into the waterside trees. Graham mentions that the most astonishing thing this time around was that the lake was 95% covered in thin ice! The bat of course had found the other 5% unfrozen water (all thanks to some very clumsy geese after they had landed on the lake).
It led Graham to ask what exactly the bats were doing – were they drinking, actively foraging, both, or neither! In truth, it is hard to say. It was certainly a very surprising encounter with this little hunter. I think this story just goes to show that we have so much to learn about bats' activities through the winter and perhaps climate change is also going to have its effects. We all hope that these two bats were able to survive the winter.
Thank you Graham White for sharing your story and photos!

Have any readers seen anything similar? I'd love to hear about it!

Monday, 15 January 2018

Mysteries of the night

Mysteries of the night by Helen Hayes MP (species champion for the common pipistrelle)

Everyone finds different ways of coping with the shorter days and colder winter weather but some British wildlife have come up with enviable methods of getting through the worst of this season. As species champion for the common pipistrelle bat I have been discovering more about how this tiny creature, which weighs less than a pound coin, copes with the harshest of winters. Their tactic is to sleep, or rather hibernate, through the worst of the weather and start afresh when the warm weather returns. I am sure many people would quite like to spend winter that way!

Helen Hayes MP at the start of her bat walk
A lot of my knowledge about this tiny flying mammal comes from a bat walk I took part in last October. The bat walk was in West Norwood Cemetery where I was lucky enough to be accompanied by Jo Ferguson who is the Built Environment Officer for the Bat Conservation Trust (BCT) and Ian Boulton, who is the Environmental Compliance Officer for the London Borough of Lambeth. Both of these ecologists provided me with an array of interesting facts about these fascinating and important animals.  All 18 UK resident bat species eat insects and in winter these are just too scarce to be a reliable source of food so all of our bats spend the winter hibernating. Once the weather warms up enough they awake to restart their nightly chase of flying insects again.

The common pipistrelle, a species I am very proud to champion, is found right across the UK including in my London constituency of Dulwich and West Norwood. It is by far the most common bat species in the UK although its numbers are thought to have declined dramatically throughout the 20th century. I am happy to report that the work of the National Bat Monitoring Programme run by BCT shows that there are some promising signs of recovery in the common pipistrelle bat and some of the other 10 bat species they are able to monitor. Bats are a great indicator of the quality of our environment, so these signs of improvement should be welcomed by all of us.

Despite being the most common bat species, it remains a mystery as to where most common pipistrelle bats hibernate. This mystery is not restricted to the UK and researchers in the Netherlands were surprised to find that large numbers of pipistrelles were hibernating in the expansion gaps in the balconies of tower blocks.  Could some of Londons pipistrelles be hibernating in tall modern buildings? I am sure with the ongoing dedication of all the passionate volunteers involved in bat conservation we may eventually find out their secret but for now much about their lives remains a mystery.

On my bat walk I learned that there are steps we can all take to make cities more welcoming for bats and other wildlife. There is growing evidence that taking this approach is not just good for them but good for us too. Do take a look at the BCT website ( to find out more about bats and how to get involved in their conservation. There are local bat groups right across the UK including the London Bat Group.

This year I will not only be looking forward to the warmer spring weather but will also be keeping an eye on the night skies so I can spot a different sign of spring, the acrobatic flights of the common pipistrelle bat.

Helen Hayes MP receiving her common pipistrelle poster from Kit Stoner (Joint CEO)

To find out more about the Species Champion Project go to: