By Emilia Decker and Dr Simon Linke
Like terrestrial sounds, the underwater sound signature of rivers has cycles – diurnal, lunar, seasonal and annual. When investigating these sounds and their relationships with the environment in a non-invasive way, one has to account for these cycles – important sounds can only occur at certain times, just like some birds only sing at specific times of the day.
In recent years ecoacoustic methods are used to monitor the state of populations and ecosystems, however many questions in sampling and analysis design – especially around temporal patterns remain unanswered.
Our new study is only the fourth study globally that comprehensively catalogued underwater sounds in a freshwater system – and the first one in the Southern Hemisphere. At Talaroo Station – an aboriginal managed Nature Reserve, we recorded two waterholes for 6 days.
We found that fish increased their singing patterns in the morning, bug and beetle chorus at night and a creek flow in the afternoon. These daily patterns could be detected using acoustic indices – similar to ecological community indices – that measure the soundscape as a whole. This is a significant result, as this is an automated method that can continuously monitor ecosystem change in a very simple way.
Though the soundscape was driven by those three acoustical events, less dominant sounds needed to be detected by manual annotation. This is often needed – frequently it is rare sounds that are of the most interest. We therefore analysed the recording by manually annotating all sounds in every 5 seconds every 10 minutes.
We annotated over 8000 sounds, including sounds from three fish species, six water bugs sounds and ten sounds that originated from aquatic beetles, though in the end we concluded that only 60 % of the recorded sound events would have been needed to ensure same results.
All in all only 10-20% of the sound events need to be annotated for most sound types to be described.
To characterise the soundscape, we found that it was most important to record for 24 hours, with the second most important factor being to record frequently (i.e. every 10 minutes). While it is ecoacoustic mantra to record for at least 1 minute, we found recording time of samples less important – with very short snippets giving us enough information to capture all sounds.
In constant conditions, we also found that three recording days are easily enough to characterise a soundscape apart. However, one of the most stunning results was that during full moon, the Hemiptera stopped singing. This is currently being investigated during a long-term recording in a freshwater creek in South East Queensland.
The aim of the study was to investigate sampling regime and analysis to find an optimal way to further explore the freshwater orchestra. While simple methods like acoustic indices can detect the major players, details have to be analysed by manual annotation.
This work was conducted by Simon Linke & Emilia Decker from Griffith University, in conjunction with Toby Gifford (formerly Queensland Conservatorium, now Monash) and Camille Desjonqueres (University of Wisconsin-Milwaukee).
We would especially like to acknowledge our research partnership with the Ewamian Aboriginal Corporation, especially Sharon Prior and the Ewamian range team who permitted us access to their land and supported us with their knowledge of the country, field vehicles, housing and amazing catering.
This work appeared in Freshwater Biology