Elsevier

Ecological Informatics

Volume 21, May 2014, Pages 4-12
Ecological Informatics

Applying bioacoustic methods for long-term monitoring of a nocturnal wetland bird

https://doi.org/10.1016/j.ecoinf.2013.12.009Get rights and content

Abstract

Bioacoustic monitoring is becoming more and more popular as a non-invasive method to study populations and communities of vocalizing animals. Acoustic pattern recognition techniques allow for automated identification of species and an estimation of species composition within ecosystems. Here we describe an approach where on the basis of long term acoustic recordings not only the occurrence of a species was documented, but where the number of vocalizing animals was also estimated. This approach allows us to follow up changes in population density and to define breeding sites in a changing environment. We present the results of five years of continuous acoustic monitoring of Eurasian bittern (Botaurus stellaris) in a recent wetland restoration area. Using a setup consisting of four four-channel recorders equipped with cardioid microphones we recorded vocal activity during entire nights. Vocalizations of bitterns were detected on the recordings by spectrogram template matching. On basis of time differences of arrival (TDOA) of the acoustic signals at different recording devices booming bitterns could be mapped using hyperbolic localization. During the study period not only changes in the number of calling birds but also changes in their spatial distribution connected with changes in habitat structure could be documented. This semi-automated approach towards monitoring birds described here could be applied to a wide range of monitoring tasks for animals with long distance vocalizations.

Introduction

Recently bioacoustic methods have become a powerful tool for monitoring biodiversity. Their specific potential lies in the detection of cryptic vocalizing animals, even in the absence of an observer. Birds are good subjects for bioacoustic census methods since most of the species use vocalizations to attract mates and to advertise territories (Gaunt and McCallum, 2004). Most of the recently used monitoring schemes for birds are based on the detection of territorially behaving animals, mostly through mapping singing males (Gregory et al., 2004). In applying bioacoustic methods for nature conservation we must bear in mind that the results should meet the requirements for long-term monitoring. We should expect an improvement in monitoring schemes either in terms of effectiveness or in terms of accuracy. The aim of monitoring biodiversity is to assess changes in ecological communities through time. Surveys should be designed in such a way that the obtained data are precise enough to identify trends (Magurran et al., 2010). Good indicators on environmental health are based on quantitative surveys allowing estimation of population trends of key species (Gregory and van Strien, 2010). This means for bioacoustic monitoring that we need an appropriate recording technique, an appropriate data acquisition protocol, effective tools to detect species within the recordings and methods for estimating the number of animals.

Recent digital audio recording devices allow unsupervised registration of vocal activity for several hours over a time schedule of even several months. For monitoring the acoustic activity of wild animals both specialized recorders and custom built solutions based on consumer recorders are now used (Fristrup and Mennitt, 2012, Mennill et al., 2012, Steer, 2010, Venier et al., 2012). Based on the sound material, the species composition of bird communities can be estimated. When using high sensitivity or highly directional microphones, by listening to sound recordings only small differences have been found in the number of species detected on the recordings and number of species detected directly at the observation sites (Haselmayer and Quinn, 2000, Hobson et al., 2002). However when microphones with low sensitivity were used, the observer at the study site obtained better results than an observer in the lab listening to the recordings (Hutto and Stutzman, 2009). Making sound recordings in parallel with a common bird census, similar results with respect to species composition were obtained using point count surveys (Venier et al., 2012) or line mapping (Frommolt et al., 2008). Wimmer et al. (2013) have demonstrated that targeted sampling with acoustic sensors could even reveal a significantly higher number of bird species than would be detected in a traditional bird survey.

In the future, species composition could be more effectively estimated by applying acoustic pattern recognition techniques. There are already some specialized recognizers allowing for automatic detection of certain species within long-term recordings (Bardeli et al., 2010, Towsey et al., 2012). Based on a linear predictive coding (LPC) transform, Boucher et al. (2012) could even distinguish all 14 species of a down chorus in a semi-rural location in Australia with a priori knowledge of species composition.

The biggest challenge seems to be determining the number of individuals on the basis of sound recordings. Stereo or quadraphonic records make it possible to reproduce a two or three-dimensional impression of the soundscape and thus facilitate an estimate of the number of birds by a skilled observer listening to the recordings (Celis-Murillo et al., 2009, Rempel et al., 2005). Another approach is the use of individual voice features in order to distinguish birds. This requires, however, that the acoustic characteristics are stable enough at least for the acquisition period. Gilbert et al. (1994) previously applied sound recordings for population surveys of Eurasian bitterns (Botaurus stellaris) and black-throated loons (Gavia arctica) to distinguish between individuals. Since 1990 sound recordings were used to monitor the (very small) British bittern population and to recognize individuals by voice (Gilbert et al., 2002).

The number of recorded calls or song bouts has also been used as a measure of the number of animals, especially when studying bird migration. However it was difficult to estimate the absolute number of animals because the call rate differs between species and even within a species, the call rate being strongly dependent on the particular environmental conditions (Farnsworth, 2005, Farnsworth et al., 2004, Hüppop and Hilgerloh, 2012). Buxton and Jones (2012) found that nocturnal call activity can at least provide information on the relative population density of cave breeding birds. Comparing different methods of bird surveys Buckland (2006) found that the census based on song activity delivered results with a rather poor precision.

Microphone arrays open new ways for monitoring birds by allowing an estimate of the location of the vocalizing animal (Blumstein et al., 2011). On basis of the time delay for the arrival of the signals at different microphones, the position of the vocalizing animal can be estimated with an accuracy of a few meters depending on the distance between the sensors (Mennill et al., 2006, Mennill et al., 2012). However Wahlberg et al. (2003) could localize booming bitterns only with an accuracy of more than 100 m using a four microphone array with a distance between neighboring microphones of between 65 and 294 m.

Here we present data from a long-term monitoring project where a quantitative assessment of a bird population was achieved on the basis of audio recordings. As a focus species we selected the Eurasian bittern. The characteristic vocalization of this species is a loud boom which is uttered in sequences of a few elements, called boom trains. The main element of the boom, which can be heard over distances of more than 1 km, is a tonal element with a mean frequency of approximately 150 Hz and a mean duration of approximately 500 ms (McGregor and Byle, 1992). In regions with high population density the population size of bitterns is difficult to estimate since the bird is hard to detect in dense reed vegetation, most vocal activity occurs at night and the signal is almost impossible to localize by a human listener. Our approach includes long-term sound recordings, detection of calls of the focus species within the recordings, discrimination of individual animals based on acoustic features and spatial distribution and a plot of a distribution map.

Section snippets

Study area

Our study was conducted in the newly created wetland restoration area “Polder Große Rosin” in the north-eastern Germany. The restoration area belongs to the Nature Reserve “Peenetal von Salem bis Jarmen”. Formerly the landscape along the river Peene was characterized by extended reed bog complexes. Due to amelioration, the area was transformed during the last century into extended grassland areas. Dramatic changes occurred in the 1970s when the area was surrounded by dams. Pumping stations were

Detection of vocalizations within long-term recordings

To prove the quality of the pattern recognition procedure we randomly selected from the long-term recordings 20 snippets of 15 min. By visual inspection of the spectrograms 497 bittern calls were detected. The template matching procedure with the sine tone recognized 422 (84.9%) of them. The evaluation of the recognizer revealed only one false positive (misinterpretation) during 300 min of recording. Therefore we could rely on a highly effective generic template to search for bittern booms.

On the

Discussion

We have shown that bioacoustic methods could be successfully applied to monitoring projects. As far as we know it is the first long term project where changes in population size and distribution were studied relying only on acoustic recordings. We should note that the case of the Eurasian bittern is a simple situation relative to the acoustic problem, since the call of the species is loud and – due to its low frequency – travels over large distances. In our study the main problems for automatic

Conclusions

Bioacoustic approaches can not only deliver data on the occurrence of certain species, but can also be used to determine the exact number of territorially behaving animals. In some cases the results of a bioacoustic census are even more reliable than conventional census methods. Using hyperbolic localization individuals can be distinguished even over large distances where voice parameters could not be reliably extracted for individual recognition. Even if we do not have fully automated

Acknowledgements

For assistance during field work we thank Benjamin Herold, Alexander Eilers, Martina Koch, Tom Lau, Anita Giermann, Judith Willkomm, Gerald Lordan and Wolfgang Wiehle. Customized adjustments of the recording equipment were conducted by Hendryk Schneider and Andreas Gnensch. We thank David Chesmore and Jason Dunlop for comments and language assistance. We thank the LUNG Neubrandenburg for supporting the work in the Nature Reserve. The study was supported by a grant from the Deutsche

References (42)

  • R. Bardeli et al.

    Detecting bird sounds in a complex acoustic environment and application to bioacoustic monitoring

    Pattern Recognit. Lett.

    (2010)
  • S.L.L. Gaunt et al.

    Birdsong and conservation

  • A.E. Magurran et al.

    Long-term datasets in biodiversity research and monitoring: assessing change in ecological communities through time

    Trends Ecol. Evol.

    (2010)
  • A.M. Ali et al.

    An empirical study of collaborative acoustic source localization

    J. Signal Process. Syst.

    (2009)
  • D.T. Blumstein et al.

    Acoustic monitoring in terrestrial environments using microphone arrays: applications, technological considerations and prospectus

    J. Appl. Ecol.

    (2011)
  • N. Boucher et al.

    A fully automatic acoustic monitor and servey system

  • S.T. Buckland

    Point-transect surveys for songbirds: robust methodologies

    Auk

    (2006)
  • R.T. Buxton et al.

    Measuring nocturnal seabird activity and status using acoustic recording devices: application for island restoration

    J. Field Ornithol.

    (2012)
  • A. Celis-Murillo et al.

    Using soundscape recordings to estimate bird species abundance, richness, and composition

    J. Field Ornithol.

    (2009)
  • O. Cramer

    The variation of the specific heat ratio and the speed of sound in air with temperature, pressure, humitidy, and CO2 concentration

    J. Acoust. Soc. Am.

    (1993)
  • A. Digby et al.

    A pracrical comparison of manual and autonomous methods for acoustic monitoring

    Methods Ecol. Evol.

    (2013)
  • A. Farnsworth

    Flight calls and their value for ornithological studies an conservation research

    Auk

    (2005)
  • A. Farnsworth et al.

    A comparison of nocturnal call counts of migrating birds and reflectivity measurements on Doppler radar

    J. Avian Biol.

    (2004)
  • K.M. Fristrup et al.

    Bioacoustical monitoring in terrestrial environments

    Acoust. Today

    (2012)
  • K.-H. Frommolt et al.

    Advantages and disadvantages of acoustic monitoring of birds – realistic scenarios for automated bioacoustic monitoring in a densely populated region

  • G. Gilbert et al.

    Vocal individuality as a census tool: practical considerations illustrated by a study of two rare species

    J. Field Ornithol.

    (1994)
  • G. Gilbert et al.

    Local annual survival of booming male Great Bittern Botaurus stellaris in Britain, in the period 1990–1999

    Ibis

    (2002)
  • R.D. Gregory et al.

    Wild bird indicators: using composite population trends of birds as measure of environmental health

    Ornithol. Sci.

    (2010)
  • R.D. Gregory et al.

    Bird census and servey techniques

  • J. Haselmayer et al.

    A comparision of point counts and sound recording as bird survey methods in Amazonian southeast Peru

    Condor

    (2000)
  • K.A. Hobson et al.

    Acoustic surveys of bird using electronic recordings. new potential from an omnidirectional microphone system

    Wildl. Soc. Bull.

    (2002)
  • Cited by (0)

    View full text