Elsevier

Ecological Indicators

Volume 13, Issue 1, February 2012, Pages 46-54
Ecological Indicators

Monitoring animal diversity using acoustic indices: Implementation in a temperate woodland

https://doi.org/10.1016/j.ecolind.2011.05.006Get rights and content

Abstract

Biodiversity assessment is one of the major challenges for ecology and conservation. With current increase of biodiversity loss during the last decades, there is an urgent need to quickly estimate biodiversity levels. This study aims at testing the validity of new biodiversity indices based on an acoustic analysis of choruses produced by animal communities. The new Acoustic Richness index (AR) and the former dissimilarity index (D) aim at assessing α and β diversity respectively. Both indices were tested in three woodland habitats: a mature forest, a young forest and a forest-cropland ecotone within the Parc Naturel Régional of Haute-Vallée de Chevreuse (France). Three recorders running for 74 days generated 5328 files of 150 s for a total of 222 h of recording. All files were treated with frequency and amplitude filters to try to remove anthropogenic and environmental background noise. The AR index was in agreement with traditional aural identification of bird species. The AR index revealed an expected gradient of diversity with higher diversity values in the young forest that potentially provides a higher number of microhabitats. The D index also indicated expected differences in the acoustic environment across sites with distinct habitat structure. Both indices reveal significant peak during dawn chorus. These results suggest that diversity could be estimated through acoustics at both α and β scales. Our analyses reveal that, even if background noise needs to be considered with great care, the use of acoustic indices has the potential to facilitate animal diversity assessments over seasons or years and landscape scales.

Introduction

In the last three decades, several indices have been developed to assess biodiversity. Most indices concerned species diversity and abundance including general density, average geometrical index or relative abundance, specific richness, Simpson or Shannon index (e.g. Margalef, 1958, MacArthur, 1965, Whittaker, 1972, May, 1975, Magurran, 2004, Buckland et al., 2005). Others indices have included characteristics of the species such as genetics, phylogenetics and functional traits (e.g. Faith, 1992, Pavoine et al., 2004, Petchey and Gaston, 2006). Interest has also been given to a simplification of these indices into α diversity, which measures the diversity within areas, and β diversity that evaluates differences among areas providing information on the turnover of specific diversity (Whittaker, 1972, Diserud and Odegaard, 2007). To evaluate changes in diversity pattern through time, both α and β diversity have to be assessed at different day or season times. Most methods require large-scale data sampling at several locations and dates. However, traditional sampling methods are mainly based on slow inventories that may not be adapted to rapid assessment at large scales in particular when dealing with highly diverse groups as arthropods (Basset et al., 2000, Lawton et al., 1998). In addition, these sampling methods are in most cases invasive as relying on direct collection or trapping (Sutherland, 1996, Hill et al., 2005). We therefore propose to adapt non-invasive acoustic analyses to quickly reveal spatial and temporal patterns of variation in animal diversity.

Acoustics can be used for animal movement tracking (Hammer and Barrett, 2001, Mennill et al., 2006), automatic species identification (Brandes et al., 2006, Brown et al., 2006, Chen and Maher, 2006, Villanueva-Rivera, 2007, Brandes, 2008, Bardeli et al., 2010), animal conservation (Laiolo, 2010), singing activity estimation (Pieretti et al., 2011) and Rapid Biodiversity Assessment (RBA) (Brandes, 2005). RBAs based on animal vocalisations can provide simple biodiversity estimations through aural identification or recording of acoustic communities (Herzog et al., 2002, Rempel et al., 2005, Villanueva-Rivera, 2007) and can also determine community structure (Diwakar and Balakrishnan, 2007a, Diwakar and Balakrishnan, 2007b, Riede, 1993, Riede, 1997). Among these RBA programs, the Automated Digital Recording System (ADRS), which allows automatic data collection, generates a large amount of high-quality data (Acevedo and Villanueva-Rivera, 2006), considerably reduces human related costs (Parker, 1991, Penman et al., 2005) and avoids any invasion of the prospected habitat (Diwakar et al., 2007).

Accordingly, a new RBA method through acoustics has been recently developed (Sueur et al., 2008b, Obrist et al., 2010). The main principle of this method is to obtain a holistic view of the local animal community that produces sound. The method avoids species or morpho-species identification, a pre-requisite of other inventory methods, All Taxa Biodiversity Inventory (ATBI) and RBAs included.

Recently an acoustic method has been proposed also to infer the singing activity of avian communities (Pieretti et al., 2011). The number of bird vocalisations produced within a community is estimated through an algorithm that computes the variability of the sound intensities. This method looks promising but cannot be used to assess species diversity and has not been employed yet to compare the acoustic composition of different communities. Based on a simple analysis of the signal, acoustic indices inferring α and β animal diversity were previously developed to summarize the variety of choruses produced by animal communities (Sueur et al., 2008b). The acoustic indices, named H and D, respectively, were first tested on artificial choruses whose specific diversity was known, and later successfully applied in situ in a coastal tropical forest of Tanzania where animal diversity is not known in detail (Sueur et al., 2008b). Here, we test and develop the indices in a temperate habitat where the diversity is lower than in Tanzania and the background noise due to human activity (anthrophony sensu Qi et al., 2008 and Pijanowski et al., 2011) may be prominent. In particular, we tested during springtime the acoustic indices in three distinct habitats: (i) a mature open forest with few vegetation strata and low tree density, (ii) a young closed forest with several vegetation strata and high tree density, and (iii) a forest-cropland ecotone with a crop field inhabited by a few species. We mainly focused our test on singing birds that are the major source of sound diversity during spring. Amphibian vocalisations were scarce and insect stridulations were notably absent. Following previous studies on bird richness in different woodland habitats (Blondel et al., 1973, Fuller and Crick, 1992, Tellería et al., 1992), a higher bird diversity can be expected in the younger forest than in the mature forest and at the forest-cropland ecotone.

Computing the acoustic H and D indices in such habitats, we estimated that local background noise can significantly impair the results provided by the H index. This led us to develop a new α index, named Acoustic Richness (AR), based on the temporal entropy and amplitude of the signal. We then addressed the following questions regarding both AR and D acoustic indices temperate habitats: (i) do the indices match with results provided by a classical bird inventory? (ii) could the indices follow the variation of daily animal activity? and (iii) could the indices highlight expected biodiversity differences between different habitats?

Section snippets

Study area and recording

Fieldwork was carried out in the Parc Naturel Régional of Vallée de Chevreuse (24,300 ha), a protected area located 40 km southwest of Paris, France. The area is deeply transformed by human activities with a land cover of 20% dwellings, 40% crop fields, and 40% temperate mixed deciduous forest.

The recording equipment consisted of three digital audio field recorders Song Meter SM1 (Wildlife Acoustics, 2009). These off-line and weatherproof recorders are equipped with an omni-directional microphone

Meteorological factors

The RDA applied to AR (matrix with sites as columns, the days as rows, and AR averaged values as entries) and meteorological parameters as factors revealed the importance of precipitation (Fig. 2). Precipitation was strongly correlated (r = 0.95) with the first axis that explained 87.5% of variation. AR mean was also positively correlated with the first axis, i.e. positively increased with precipitations and, to a lesser extent, wind. The first canonical plan showed an opposition between bad

Discussion and conclusions

Animal diversity is traditionally estimated with species inventories. However, such inventories rely on a series of difficult tasks. In most cases, specimens have first to be collected in the field. Once brought back in a Museum collection, specimens have to be sorted out, prepared for examination, and eventually each specimen needs to be identified by a taxonomic expert. Alternatively, vocalising species can be estimated in the fields through aural identification with trained human listeners (

Acknowledgements

We would like to thank François Hardy and Grégory Patek (PNR of Haute Vallée de Chevreuse) for their help during data acquisition. We are indebted to Mathieu Benguigui for his computer assistance. We are grateful to Tristan Tyrrell (UNEP-WCMC, Cambridge) for having improved the language in the manuscript. This study was funded by the Institut Français de la Biodiversité (’Jeunes Chercheurs’ grant) and the Fondation pour la Recherche sur la Biodiversité (’BIOSOUND’ grant). We thank two anonymous

References (60)

  • Brandes, T.S., 2005. Acoustic Monitoring Protocol. Tropical Ecology Assessment and Monitoring (TEAM) Initiative Set of...
  • T.S. Brandes

    Automated sound recording and analysis techniques for bird surveys and conservation

    Bird Conservation International

    (2008)
  • T.S. Brandes et al.

    Using image processing to detect and classify narrow-band cricket and frog calls

    Journal of the Acoustical Society of America

    (2006)
  • A.S. Bridges et al.

    Temporal variation in anuran calling behavior: implications for surveys and monitoring programs

    Copeia

    (2000)
  • T.J. Brown et al.

    Why birds sing at dawn: the role of consistent song transmission

    Ibis

    (2003)
  • J.C. Brown et al.

    Classification of vocalizations of killer whales using dynamic time warping

    Journal of the Acoustical Society of America

    (2006)
  • S.T. Buckland et al.

    Monitoring change in biodiversity through composite indices

    Philosophical Transactions of the Royal Society B—Biological Sciences

    (2005)
  • Z. Chen et al.

    Semi-automatic classification of bird vocalizations using spectral peak tracks

    Journal of the Acoustical Society of America

    (2006)
  • W.S. Cleveland

    LOWESS: a program for smoothing scatterplots by robust locally weighted regression

    American Statistician

    (1981)
  • R Development Core Team, 2010. The R Project for Statistical Computing. http://www.r-project.org/ (accessed...
  • V. Devictor et al.

    Birds are tracking climate warning, but not fast enough

    Proceedings of the Royal Society, Biological Sciences B

    (2008)
  • O.H. Diserud et al.

    A multiple-site similarity measure

    Biology Letters

    (2007)
  • S. Diwakar et al.

    The assemblage of acoustically communicating crickets of a tropical evergreen forest in southern India: call diversity and diel calling patterns

    Bioacoustics

    (2007)
  • S. Diwakar et al.

    Vertical stratification in an acoustically communicating ensiferan assemblage of a tropical evergreen forest in southern India

    Journal of Tropical Ecology

    (2007)
  • S. Diwakar et al.

    Psychoacoustic sampling as a reliable, non-invasive method to monitor orthopteran species diversity in tropical forests

    Biodiversity and Conservation

    (2007)
  • S. Dray et al.

    The ade4 package: implementing the duality diagram for ecologists

    Journal of Statistical Software

    (2007)
  • R.J. Fuller et al.

    Broad-scale patterns in geographical and habitat distribution of migrant and resident passerines in Britain and Ireland

    Ibis

    (1992)
  • N.J. Gotelli et al.

    Null Models in Ecology

    (1996)
  • O. Hammer et al.

    Techniques for studying the spatio-temporal distribution of animal vocalizations in tropical wet forests

    Bioacoustics

    (2001)
  • S.K. Herzog et al.

    Estimating species richness of tropical bird communities from rapid assessment data

    Auk

    (2002)
  • Cited by (265)

    View all citing articles on Scopus
    View full text