If a tree falls in the forest, does it make a sound?

A recent study led by myself and Dr. Nick Friedman asks whether we can accurately measure how diverse different ecosystems are on the island of Okinawa, Japan. We set up 24 monitoring sites across the island in different locations – in forests, grassland, mangroves, near the beach and in the city – to monitor all the sounds that are produced near each site. We found that we can detect individual species and relate these sounds to natural patterns including the ‘dawn chorus,’ and we could identify sites with heavy human activity. All without having to look for any species.

Me (with full ecology beard) pinning an acoustic monitor to a tree in Okinawa.

The rise of bioacoustics

Technology is advancing worldwide. Everything from phones to microwaves is getting more advanced. Instruments for ecological research are no different. Our satellite tracking tags are improving; they’re getting lighter, cheaper and can store more data than ever before. We can use complex chemical techniques to understand who eats whom in a food web, and drones now allow us to image even remote habitats with relative ease. With these advances comes the rise of acoustic monitoring techniques for biological signals (bioacoustics for short).

Bioacoustic monitoring relies on a recording device which is usually left in the field to record for several days, weeks, months or even years. The devices are basically microphones that record everything they hear either continuously or in a regular pattern. Then the data from the devices can be analysed by ecologists in a lab or office in their own time. This means that bioacoustic data doesn’t have to be collected manually like traditional monitoring techniques for biodiversity (e.g. plant or bird surveys) – a huge advantage over regular sampling techniques.

Identifying species

We can use bioacoustic data to identify species by listening to the recordings from devices deployed in the field. If we hear an individual species, we can use that data to get information on the presence of that species at a particular time and place. Using software anyone can download on a standard computer, we can also look at the ‘waveform’ of a recording and try to identify specific characteristic sounds based on the patterns in the sound structure of the recording.

Some species can be easier to detect than others (see this previous post about my difficulties with Corvus macrorhynchos), and it’s down to us to decide whether this is due to error with the recordings or analysis, or whether it’s a true pattern of species abundance; some species are more common than others and/or vocalise more often and so will be easier to hear on recordings than quiet or rare species.

Female Blue Rock Thrush (Monticola solitarius) in Okinawa – a species we can detect but was not a focal species in our study.


Listening to everything: the soundscape approach

Another way to monitor trends in biodiversity using acoustic monitoring is to listen to everything on a recording and then condense it all into one metric. We call this approach ‘soundscape ecology’ because it recognises that it’s not just animals that contribute to the total range of sounds recorded in an ecosystem (i.e. it’s soundscape). Other sounds include human-related noise such as traffic, construction noise or dogs barking (plus people talking, shouting and laughing!) and other natural sounds that aren’t animals such as wind, running water and branches breaking.

Soundscape ecology is a growing field which shows promise due to its simplicity – it’s much faster to measure one value that represents all sound than identifying species individually. Naturally, people question how useful these data are; if we condense all biodiversity and other non-target sounds into one measure, that measure inherently contains a lot of unwanted ‘noise’ (pun slightly intended). But even if we do capture some unwanted data, soundscape ecology approaches are particularly useful for looking at general patterns, such as changes in sound composition over time (e.g. time of day or across seasons).

The OKEON-Churamori project: monitoring Okinawa’s soundscapes

Run by Dr. Evan Economo at the Okinawa Institute of Science and Technology, the OKEON-Churamori project has been established to monitor the biodiversity and environment of habitats across the island. This Okinawa Environmental Observation Network (hence OKEON) includes a network of acoustic monitors at 24 sites across Okinawa. We used these monitors to ask whether we could identify several species of birds at each of 5 key sites and measure the whole soundscape using a range of common soundscape ecology techniques.

We identified five species that are important in Okinawa’s ecosystems: the large-billed crow (Corvus macrorhynchos), the brown-eared bulbul (Hypsipetes amaurotis), the ruddy kingfisher (Halcyon coromanda), the Ryūkyū scops owl (Otus elegans) and the endangered Okinawa Rail (Gallirallus okinawae). Together these species are supposed to be a representative sample of the kinds of species we might try to detect; a few common and loud vocalising species (the crow and bulbul), a highly seasonal visitor species (Kingfisher), a nocturnal species (owl) and a rare species found only at the northern-most part of the island (the Okinawa rail).

I’ve never seen one in the wild so this Ruddy Kingfisher at Ueno Zoo will have to suffice.

Tying things together: the future of bioacoustic monitoring on Okinawa 

We managed to identify all our focal species at the locations and times we expected to find them. For example, the owl wasn’t recorded during the day, and the Okinawa Rail was only recorded at our most northern site. Our soundscape approach also identified a clear ‘dawn chorus’ between 5am and 7am each day, which is usually attributed to a peak in birdsong around this time. Our sites that were nearest the city generally had less birdsong and different soundscape values to our sites that were mainly surrounded by forest. We believe this shows an impact of urban noise and land-use change on the soundscape of Okinawa.

Perhaps most importantly, we were able to relate the dawn chorus effect that we identified through our soundscape approach to several of our focal species; particularly the ruddy kingfisher. To our knowledge, this is the first time anyone has empirically been able to identify a relationship between individual species vocalisations and patterns in the soundscape of an ecosystem such as a dawn chorus effect.

Now that we know we can accurately detect individual species and measure soundscape metrics at the OKEON sites, we can expand this study far beyond this initial small-scale trial. We can start to ask more complicated questions using more sites, different species and over longer periods of time. So, if a tree falls in an Okinawan forest, we’ll know whether it makes a sound – proving once again that ecology trumps philosophy.

To find out more, read our Ecological Research article ‘Listening to ecosystems: data rich acoustic monitoring through landscape-scale sensor networks’.

If you’re interested in working with us on these projects please get in touch with either myself or Nick Friedman .

In the news:
– Springer
– EurekAlert!
– Alphagalileo
– Science Newsline
– Phys.org
– OIST and promotional video

Paper reference:
Ross SRP-J*, Friedman NR*, Dudley KL, Yoshimura M, Yoshida T, Economo EP. (In press). Listening to ecosystems: data rich acoustic monitoring through landscape-scale sensor networks. Ecological Research. DOI: 10.1007/s11284-017-1509-5



4 thoughts on “If a tree falls in the forest, does it make a sound?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s