Bioacoustic Monitoring of Biodiversity - our first presentation of bioacoustic monitoring in cooperation with the Global Canopy Initiative (Andrew Mitchell?), in 2002
2. Mission
Develop a new method of measuring
biodiversity in the worlds forests
based on animal vocalisations
3. Introduction
• With 10’s of thousands of plant and animal species
in tropical forests, very few specialists can identify
them
• Using taxonomists to do this is expensive and time
consuming
• We propose a radically new method of assessing
biodiversity by analysing the spectral of insect
sounds in forests if successful identifying the
‘biodiverstiy hotspots’ of the world would be as
simple as walking into it with a microphone and
recording its sounds
5. Justification
• Insect diversity in forests is overwhelming
• The diversity of sounds they make is related to the diversity of
species within a forest
• No-one has developed a means of measuring forest biodiversity in
a simple, non-invasive way and highly skilled taxonomists whose
work is both expensive and time consuming
• The stereotype nature of insect sounds is specially suited for
computer analysis - providing a perfect indicator group to work on
• If such a methods could be perfected it may be possible for non
specialists to monitor biodiversity and indicate the value of
different forests simply by the way they sound
• This could significantly speed up the process of biodiversity
assessment and protection of “biodiversity hotspots”
• Technology is cheap
6. Advantages for Science
• Detect areas with a high number of un-
described species
• Characterise the mosaic of endemisms in
larger forest tracts
• Interesting theoretical aspects for
Informatics (especially speech and pattern
recognition)
7. Project development
• Identify partners in bioacoustics, informatics,
neural networks and digital signal processing
• Develop protocols - peer-reviewed by Bio-
acoustitians and ecologists
• Identify 3 tropical forest sites for calibration
• Identify 5-10 sites to initiate monitoring
• Implement field trials alongside traditional
biodiversity assessment projects
• Liase with traditional taxonomists for the
identification of species
8. Phase I - Pilot Testing of BAMBi
• Gather insect sounds by simple sound recording using
microphones at different seasons, times of day and
different heights of the forest, within selected study sites
• Measure sound prior to and in combination with traditional
collecting techniques
• Sound recording of captured individual insects and
identification of voucher specimens
• Extending sound recording to more sophisticated schemes
(ultrasound, multi-channel and hard disc recording)
• Field test neural network informatic tools
• Collect sound data alongside insect traditional collecting
techniques in the forest
9. Phase II - Expanded Field Testing
• Expand research to 7 more tropical sites
• Not necessarily canopy sites
• After development of automated analysis
– Compare pristine and disturbed sites
– Compare managed and unmanaged sites
– Compare certified and uncertified forests
10. Potential field locations
• COPAS - French Guiana
• Smithsonian Canopy Cranes - Panama
• ACCER Walkways - Iquitos, Peru
• Daintree Canopy Crane - Cairns Australia
• Lambir Hills Canopy Crane - Malaysia
• Dannum Valley, Malaysia
• Kuala Belalong - Brunei
• Kenya East Africa (BIOLOG)
11. Analysis
• Analysis of taped sounds by
traditional methods - spectral
analysis
• Analysis of taped sounds with
automated tools (neural networks)
• Databasing of individual recordings
for the virtual museum (identified
sound + voucher specimen)
12. Meta Analysis
• Creating the BAMBi index
– Generate a number of species for each site based on the
sound analysis
– Investigate variation between height, time of day, sites
and seasons
• Compare validity of Index with fogging data
• Feed sound into the Neural network and compare
the output with traditional biodiversity
assessement
13. Schedule
2002
• Development of collaborations and protocols
• Initial field trials (pilot study) - Autumn 2002
2003 - 4
• Initiate full field testing (3 sites)
2005
• Analysis and interpretation
2006
• Full project implementation
14. Personnel
BAMBi Project Supervisor - Dr Klaus Riede - Bonn University,
Gemany
Field Testing
• 2002 field test - Frank Nischk, Professor Heiner Roemer (STRI, Panama)
• Full time pHD x 1 (3 yrs)
• Field assistant x 1(3 months / yr)
Taxanomic Data Processing
• Lab assistant for sorting (3 mths)
• Specialist Taxonomist (6 mths)
Digital Data Processing
• Digital Sound Processing, Professor Mark Sandler & Dr Mark Plumbly QMW, London
• Professor Guenther Palm, Neuro-Informatics, Ulm University
Social Value Study
• Professor David Pearce, University College London
• Andrew Mitchell, Green College Oxford
16. Applications of BAMBi
• Biodiversity Assessment
– Easy to use / non- invasive / rapid assessment technique
• Monitoring
– Repeatable assessments to assess biodiversity change
– Monitoring and mapping of indicator species
• Forest management
– Assessing impacts of logging and forest conversion
– Impact of different management regimes (plantations)
• Virtual Museum
– Databasing and web access to library of wildlife sounds
– Archiving today’s forest sounds prior to extinction?
• Economics
– Placing a social value on biodiversity
– Aesthetics of wildlife sounds
– Spin offs for automatic speech recognition and music industriesr
• Policy
– Informing decision makers / stakeholders
– Labelling and certification schemes for tropical economies
– Creation of tradable certificates based on biodiversity
• Capacity Building
– Training of non specialists to monitor forests
18. Budget
Per year
€
3 yr
€
Total
€
Personnel
pHD x2
Assistants x2
Taxonomist
50,000
6,660
8,333
150,000
20,000
25,000
195,000
Equipment 20,000 20,000
Travel (3 field trips x2 prs)
3,333 10,000 10,000
Subsistence (2 prs x30 days x 3 field trips @50€/day)
3.000 9.000 9.000
Taxonomy Overhead 1,667 5,000 5,000
Digital Analysis 25,000 75,000 75,000
Social Value Study 25,000 75,000 75,000
Coordination & Management 15% total b 58,350
TOTAL 447,350