understanding connectivity through song

 

Our previous work on Shortwing song showed that two genetically different sky islands were different in breeding song structure (See Shortwing Song section). This song project extends this work to examine differences in song across all sky islands in the species’ range. Here we examine if populations isolated by historic isolation differ in aspects of songs that are different from populations isolated by recent deforestation. We also correlate song differences with population connectivity based on genetics. Watch this space for more updates.

SHORTWING song across islands

Ramki Sreenivasan

Funding

 


                                         

IIT-Chennai



Support


        

  

NIAS

          

NCF



 

Automated song monitoring

Shola habitat on any large sky island is fragmented and disconnected by the formation of commercial plantations about 100 - 150 years ago. This production landscape now hosts remnant shola forest patches that have been reported to harbour many threatened, endemic birds including the Shortwing. However, it is not clear if these birds are passing through these patches or if they are using these over a longer term for breeding.


This collaborative project is with Anil Prabhakar, from IIT Chennai, an avid birdwatcher who has always been interested in avian acoustics and has experience in signal processing. He has also been working on creating several automated devices and now hopes to build an automated recording device that will send us recordings of birds using the cellular phone network from this sky island landscape. The IIT Chennai team also includes Nitin Chandrachoodan helping us build the embedded systems and the wireless network and includes a few rotating students.


This collaborative study uses technology that can be used to answer several biological questions.  We propose to examine occupancy of remnant forest patches by threatened, endemic birds in a production landscape. This project will provide the owners and managers of production landscapes in priority sites with an assessment of avian biodiversity in their land holdings. This information may be useful for these managers to get their private holdings audited or certified for green and conservation practices. The project will also engage with these owners and managers to sensitise them on the importance of such species. Over a longer period this can also lead to an understanding of timing of breeding of birds in the sky island complex and any possible effect of climate change.


2012:  A simple SEAV model was able to distinguish and identify two bird species. Mahesh Nandwana wins a poster presentation award for this study at a student conference, YETI Guwahati.


2013: We were able to build an algorithm, a combination of SEAV and HMM models, that could identify seven species of birds that we had good training data for.


2014: We realized that to scale up to more species, we require way more training data than what we could generate. The project has now grown to include more collaborators - Samira Agnihotri who worked on mimicry of bird song and automated algorithms to identify these is now part of the project, Tarsh Thaekekara who works on biodiversity monitoring using local communities is also part of the project. We have also moved the project to a largely citizen science model where a broader community can contribute songs (training data) that can eventually be used to build a better algorithm. This project now has its own website www.bioacoustics.in. Further updates will be on that site.


2016: This project is now driven as collaboration of three distinct components:

  1. A)Software - This involves developing an algorithm to identify individual bird species. This is led by Dr. Padmanabhan Rajan from IIT-Mandi

  2. B)Hardware - Involved developing hardware to record songs automatically from different locations. This is led by  Vaibhav Pratap Singh

  3. C)Biology - involves deploying recorders, curating birds songs, identifying and checking accuracy of classification. This is led by V.V. Robin & Samira Agnihotri

Ramki Sreenivasan

  



Kailash holds a Masters degree in Communication and has been part of this project from June 2012 to May 2013. His interests range from evolution of human communication, animal communication, non verbal codes and media studies. He helped put up recorders on the different sky islands. In this process he figured he has an amazing ability to climb hills, running up these mountains and breaking various local records! 




Suma is a final year undergraduate student from Bangalore who was awarded the Summer Research Fellowship by IAS-INSA-NASc during the year 2012 where she worked on the song of Oriental Magpie Robin in Haridwar. Recipient of the INSPIRE Scholarship from DST, She has now started working on the song of Malabar Whistling Thrush.



 

Kailash Koushik

(2012 -2013)


Suma M.

(2012 -2013)




Savini is currently doing her M.E in Communication Systems at B.I.T.S Pilani. 

She is working on a bird song recognition algorithm with the objective to improve efficiency and accuracy of species identification. This will help reduce manual efforts behind analyzing bird songs.

Excited about helping with this cross over between Engineering and Biological Science. 





Mahesh joined the project initially as a  final year undergraduate student from the Department of Electrical and Electronics Engineering NIT Jamshedpur. Later he came back to work on this project after his graduation. As a Summer Research Fellow 2011 at IIT Madras, Mahesh developed a new algorithm for bird call identification based on speaker identification that won the 3rd poster  prize at Young Ecologist Talk & Interact 2011. Mahesh is interested in Speech Signal Processing, Speech Recognition & DSP. In his second stint, he has been working on segmenting songs and extension of his algorithm to more species.

 

People in the project - PAST

Mahesh Nandwana

(2011, 2012)

  



Chetana  joined this project after working with Agumbe Rainforest Research Station. She has recorded songs on different sky islands over one year and contributed to building an extensive song library for the project. She has now moved on to the Masters programme in NCBS.

 

Chetana Purushotham

(2011-2012)




Pavan was part of the IIT team that works on the technology front of this project. His major interest is in signal processing, particularly in speech recognition. He did some preliminary work with a code to automatically separate Shortwing songs from other species.

 

Pavan Kumar

(2011)