July 21st, 2013 by Michael Ricciardi
(Note to the Reader: The following article details a new remote monitoring system for tracking animal diversity and identifying threatened species, and, as an inspirational supplement, presents a TED Talk video by Bernie Krause on the value of listening to Nature’s many sounds.)
Accurately Determining the Presence or Absence of Forest Fauna
If you’re an ecologist studying biodiversity (or its decline) in a remote habitat, there are few options for doing so other than being in the field for prolonged periods and physically observing and recording what is seen and/or heard, then long hours analyzing this information back at that lab and making educated guesses about what species one has observed (and there are relatively few field researchers that can correctly and confidently identify more than a few species based solely on their vocalizations).
This conventional approach to species monitoring is costly, limited in space and time, and, often does not result in a permanent record.
A further challenge for this vital work is that in most biomes, data collection on fauna tends to be concentrated in a few places (this is often a function of field teams of limited size, or, having only limited access to remote sites) which results in a “highly aggregated distribution of information”. This in turn limits scientists’ ability to understand the large-scale ecology of a site and thus also the ability to properly manage and protect various animals over these large areas.
A New Tool for Listening to Animal Sounds and Identifying What’s There…or Isn’t
But now, a new system for remote monitoring of natural sounds has been developed by a collaborative team from the Departments of Biology and Computer Science at University of Puerto Rico-Rio Piedras, San Juan, Puerto Rico.
The “bioaccoustic” monitoring system is called the Automated Remote Biodiversity Monitoring Network (ARBIMON) and it may very well represent a game-changer in the critical mission of cataloging biodiversity in the world’s dwindling forests and identifying endangered and disappearing species — like any of the great number of frog species and other amphibians under grave threat from climate impacts, habitat loss, and a lethal fungus invasion).
The research team that developed the system describe it as “a novel combination of hardware and software for automating data acquisition, data management, and species identification based on audio recordings.”
According tot he research paper abstract, the major components of the ARBIMON system include:
– a solar powered remote monitoring station that sends 1-min recordings every 10 min to a base station (which then relays the recordings in real-time to a dedicated server for processing. Processed recordings are then uploaded to the project server at arbimon.net).
– a module for viewing, listening, and annotating recordings; an accompanying website includes a “species identification interface” which aids users in constructing specialized algorithms to automate (and accelerate) species identification.
Predictions by the machine models were compared to the human experts’ determinations of species. Accuracy and precision of vocalization identification varied depending upon the type of animal species (e.g., predictions of larger vertebrates had higher accuracy rates than insect predictions) but with the lowest accuracy prediction dropping to no less than 76% (of the time). Although a few animal acoustic models (e.g., like the golden-mantled howling monkey Alouatta palliata) produced predictions with a “high level of false positives”, overall, the researchers found that “most of the models had relatively low levels of false positives” (<5%)” but with “higher levels of false negatives” — which means that the models predicted/reported that a species was absent (e.g., the Great Tinamou [Tinamus majoror], or Mountain hen, not currently threatened) when it was actually present (a false negative) at a higher rate (41 times, for the latter results, verses 1 time for the former) than reporting it present when it was actually absent.
According to the researchers:
‘These results suggest that these models are relatively conservative; they rarely confused the species with another, but they do not always detect the species when it is present as determined by an expert through visual and/or aural inspection.’
Analyzing thousands of lengthy field recordings to detect or extract presence/absence data requires considerable data management and analysis ability which not all researchers possess As a result, researchers tend to listen to only a subset of their total recordings. This is still time-consuming and also means less data is analyzed. The advantage to the ARBIMON system is that it reduces the time spent analyzing recordings (due to a data compression technology) while still taking advantage of the full data set
And, by compiling a permanent record of vocalizations for a given site and providing (real-time) updates, scientists will be able to compare current recordings (and their “cast” of noise-making animals) with previous ones — allowing identification of progressively disappearing sounds from some animals, over time, which could indicate a vanishing species.
The researchers demonstrated the value and utility of their technology by presenting “the vocal activity patterns of birds, frogs, insects, and mammals from Puerto Rico and Costa Rica.”
For more details on the workings of the ARBIMON system, check out the PeerJ research paper: Real-time bioacoustics monitoring and automated species identification. (researchers included: T. Mitchell Aide, Carlos Corrada-Bravo, Marconi Campos-Cerqueira, Carlos Milan, Giovany Vega, and Rafael Alvarez
TED Talk – Bernie Krause: The Voice of the Natural World
To wrap up the theme of this blog post, I offer you this related, inspirational talk by legendary recording and ‘soundscape’ engineer Bernie Krause whose past work includes work with rock group The Byrds, pop artist Stevie Wonder, and many others).
Krause has been making his nature soundscapes — of wind-rustled trees, chirping birds, and even the barely audible sounds of insect larvae — for some 45 years. His work and new TED talk will surely revive your Nature spirit — short of spending time in Nature yourself.
If this video (below) does not play for you, visit this TED.com page (and see below for a link to more of Bernie’s work!).
If you would like to hear much more of Bernie Krause’s amazing and important work, check out his BIOPHONY project (http://www.wildsanctuary.com)
Keep up to date with all the most interesting green news on the planet by subscribing to our (free) Planetsave newsletter.