Multi-Index Ecoacoustics Analysis for Terrestrial Soundscapes: A New Semi-Automated Approach Using Time-Series Motif Discovery and Random Forest Classification

Multi-Index Ecoacoustics Analysis for Terrestrial Soundscapes: A New Semi-Automated Approach Using Time-Series Motif Discovery and Random Forest Classification Marina D. A. Scarpelli Benoit Liquet David Tucker Susan Fuller Paul Roe

High rates of biodiversity loss caused by human-induced changes in the environment require new methods for large scale fauna monitoring and data analysis. While ecoacoustic monitoring is increasingly being used and shows promise, analysis and interpretation of the big data produced remains a challenge. Computer-generated acoustic indices potentially provide a biologically meaningful summary of sound, however, temporal autocorrelation, difficulties in statistical analysis of multi-index data and lack of consistency or transferability in different terrestrial environments have hindered the application of those indices in different contexts. To address these issues we investigate the use of time-series motif discovery and random forest classification of multi-indices through two case studies. We use a semi-automated workflow combining time-series motif discovery and random forest classification of multi-index (acoustic complexity, temporal entropy, and events per second) data to categorize sounds in unfiltered recordings according to the main source of sound present (birds, insects, geophony). Our approach showed more than 70% accuracy in label assignment in both datasets. The categories assigned were broad, but we believe this is a great improvement on traditional single index analysis of environmental recordings as we can now give ecological meaning to recordings in a semi-automated way that does not require expert knowledge and manual validation is only necessary for a small subset of the data. Furthermore, temporal autocorrelation, which is largely ignored by researchers, has been effectively eliminated through the time-series motif discovery technique applied here for the first time to ecoacoustic data. We expect that our approach will greatly assist researchers in the future as it will allow large datasets to be rapidly processed and labeled, enabling the screening of recordings for undesired sounds, such as wind, or target biophony (insects and birds) for biodiversity monitoring or bioacoustics research.
12 17 2021 738537 10.3389/fevo.2021.738537 1 10.3389/crossmark-policy frontiersin.org true https://creativecommons.org/licenses/by/4.0/ 10.3389/fevo.2021.738537 https://www.frontiersin.org/articles/10.3389/fevo.2021.738537/full https://www.frontiersin.org/articles/10.3389/fevo.2021.738537/full Remote Sens. Aide 9 2017 Species richness (of insects) drives the use of acoustic space in the tropics. 10.3390/rs9111096 Wavelets: Functions for Computing Wavelet Filters, Wavelet Transforms and Multiresolution Analyses. Aldrich 2020 Graphical Climate Statistics for Australian Locations. 2020 Remote Sens. Bonthoux 10 2018 Spatial and temporal dependency of NDVI satellite imagery in predicting bird diversity over france. 10.3390/rs10071136 Ecol. Indic. Bradfer-Lawrence 115 2020 Rapid assessment of avian species richness and abundance using acoustic indices. 10.1016/j.ecolind.2020.106400 Ecol. Evol. Bradfer-Lawrence 10 1796 2019 Guidelines for the use of acoustic indices in environmental research. Methods. 10.1111/2041-210x.13254 Mach. Learn. Breiman 45 5 2001 Random forests. 10.1201/9780429469275-8 Ecol. Indic. Brodie 119 2020 Automated species identification of frog choruses in environmental recordings using acoustic indices. 10.1016/j.ecolind.2020.106852 Appl. Soft Comput. J. Brown 81 2019 Automatic rain and cicada chorus filtering of bird acoustic data. 10.1016/j.asoc.2019.105501 J. Ecoacoustics Buxton 2 2018 Acoustic indices as rapid indicators of avian diversity in different land-use types in an Indian biodiversity hotspot. 10.22261/jea.gwpzvd Nature Cardinale 486 59 2012 Biodiversity loss and its impact on humanity. 10.1038/nature11148 Sci. Total Environ. Carruthers-Jones 695 2019 The call of the wild: investigating the potential for ecoacoustic methods in mapping wilderness areas. 10.1016/j.scitotenv.2019.133797 Ensemble Machine Learning Cutler 157 2012 Random forests 10.1007/978-1-4419-9326-7_5 BMC Bioinformatics Díaz-Uriarte 7 2006 Gene selection and classification of microarray data using random forest. 10.1186/1471-2105-7-3 Ecol. Indic. Doohan 96 739 2019 The sound of management: acoustic monitoring for agricultural industries. 10.1016/j.ecolind.2018.09.029 Sci. Total Environ. Duarte 769 2021 Changes on soundscapes reveal impacts of wildfires in the fauna of a Brazilian savanna. 10.1016/j.scitotenv.2021.144988 Biol. Conserv. Duarte 191 623 2015 The impact of noise from open-cast mining on Atlantic forest biophony. 10.1016/j.biocon.2015.08.006 Revision of the Interim Biogeographic Regionalisation for Australia (IBRA) and Development of Version 5.1 Summary Report. 2000 Soundscape Ecology: Principles, Patterns, Methods and Applications. Farina 2014 10.1007/978-94-007-7374-5 J. Ecoacoustics Ferreira 2 2018 What do insects, anurans, birds, and mammals have to say about soundscape indices in a tropical savanna. 10.22261/JEA.PVH6YZ For. Ecol. Manage. Fontúrbel 479 2021 Habitat disturbance can alter forest understory bird activity patterns: a regional-scale assessment with camera-traps. 10.1016/j.foreco.2020.118618 Ecol. Indic. Francomano 112 2020 Biogeographical and analytical implications of temporal variability in geographically diverse soundscapes. 10.1016/j.ecolind.2019.105845 Ecol. Evol. Froidevaux 4 4690 2014 Optimizing passive acoustic sampling of bats in forests. 10.1002/ece3.1296 Landsc. Ecol. Furumo 34 911 2019 Using soundscapes to assess biodiversity in Neotropical oil palm landscapes. 10.1007/s10980-019-00815-w Ecol. Inform. Gage 21 100 2014 Visualization of temporal change in soundscape power of a Michigan lake habitat over a 4-year period. 10.1016/j.ecoinf.2013.11.004 Ecol. Inform. Gan 60 2020 Data selection in frog chorusing recognition with acoustic indices. 10.1016/j.ecoinf.2020.101160 Proceedings of the IEEE International Conference on Data Mining (ICDM) 2017-Novem Gao 1213 2017 Efficient discovery of time series motifs with large length range in million scale time series 10.1109/ICDM.2017.8356939 Nature Gaston 405 220 2000 Global patterns in biodiversity. 10.1038/35012228 Pattern Recognit. Lett. Genuer 31 2225 2010 Variable selection using random forests 10.1016/j.patrec.2010.03.014 Methods Ecol. Evol. Gibb 10 169 2019 Emerging opportunities and challenges for passive acoustics in ecological assessment and monitoring. 10.1111/2041-210X.13101 Ecol. Indic. Hayashi 112 2020 Acoustic dissimilarities between an oil palm plantation and surrounding forests: analysis of index time series for beta-diversity in South Sumatra, Indonesia. 10.1016/j.ecolind.2020.106086 Freshw. Biol. Indraswari 65 142 2020 Assessing the value of acoustic indices to distinguish species and quantify activity: a case study using frogs. 10.1111/fwb.13222 Science Johnson 356 270 2017 Biodiversity losses and conservation responses in the Anthropocene. 10.1126/science.aam9317 Science Joppa 352 416 2016 Filling in biodiversity threat gaps. 10.1126/science.aaf3565 Ecol. Indic. Jorge 91 71 2018 The effectiveness of acoustic indices for forest monitoring in Atlantic rainforest fragments. 10.1016/j.ecolind.2018.04.001 Trends Ecol. Evol. Kerr 18 299 2003 From space to species: ecological applications for remote sensing. 10.1016/S0169-5347(03)00071-5 Biol. Conserv. Krause 195 245 2016 Using ecoacoustic methods to survey the impacts of climate change on biodiversity. 10.1016/j.biocon.2016.01.013 Bull. - Am. Meteorol. Soc. Lau 76 2391 1995 10.1175/1520-0477(1995)076<2391:CSDUWT>2.0.CO;2 Climate signal detection using wavelet transform: how to make a time series sing. R News Liaw 2 18 2003 Classification and regression by random forest. IEEE Trans. Ind. Informatics Liu 11 583 2015 Efficient motif discovery for large-scale time series in healthcare. 10.1109/TII.2015.2411226 Landsc. Urban Plan. Machado 162 36 2017 Do acoustic indices reflect the characteristics of bird communities in the savannas of Central Brazil? 10.1016/j.landurbplan.2017.01.014 Data Min. Knowl. Discov. McGovern 22 232 2011 Identifying predictive multi-dimensional time series motifs: an application to severe weather prediction. 10.1007/s10618-010-0193-7 Methods Ecol. Evol. Metcalf 2020 421 2020 Acoustic indices perform better when applied at ecologically meaningful time and frequency scales. 10.1111/2041-210x.13521 Ecol. Indic. Mitchell 119 2020 Spatial replication and habitat context matters for assessments of tropical biodiversity using acoustic indices. 10.1016/j.ecolind.2020.106717 Ecol. Indic. Moreno-Gómez 103 1 2019 Evaluating acoustic indices in the Valdivian rainforest, a biodiversity hotspot in South America. 10.1016/j.ecolind.2019.03.024 Eur. J. Ecol. Nowak 4 56 2019 Unmanned Aerial Vehicles (UAVs) in environmental biology: a review. 10.2478/eje-2018-0012 Int. J. Remote Sens. Pal 26 217 2005 Random forest classifier for remote sensing classification. 10.1080/01431160412331269698 Methods Ecol. Evol. Petrusková 7 274 2016 Repertoire-based individual acoustic monitoring of a migratory passerine bird with complex song as an efficient tool for tracking territorial dynamics and annual return rates. 10.1111/2041-210X.12496 The Normalized Difference Vegetation Index. Pettorelli 2013 10.1093/acprof:osobl/9780199693160.001.0001 PLoS One Phillips 13 2018 Revealing the ecological content of long-duration audio-recordings of the environment through clustering and visualisation. 10.1371/journal.pone.0193345 Ecol. Indic. Pieretti 11 868 2011 A new methodology to infer the singing activity of an avian community: the acoustic complexity index (ACI). 10.1016/j.ecolind.2010.11.005 Bioscience Pijanowski 61 203 2011 Soundscape ecology: the science of sound in the landscape. 10.1525/bio.2011.61.3.6 Methods Ecol. Evol. Roe 2021 1 2021 The Australian acoustic observatory. 10.1111/2041-210X.13660 Remote Sens. Ecol. Conserv. Sánchez-Giraldo 6 248 2020 Ecoacoustics in the rain: understanding acoustic indices under the most common geophonic source in tropical rainforests. 10.1002/rse2.162 Release for Multi-Index Ecoacoustics Analysis Scarpelli 2021 The Mathematical Theory of Communication. Shannon 1964 10.1109/TMAG.1987.1065451 Entomol. Exp. Appl. Stafford 130 113 2009 Characterization and correlation of DC electrical penetration graph waveforms with feeding behavior of beet leafhopper, Circulifer tenellus. 10.1111/j.1570-7458.2008.00812.x ACTA Acust. United With Acust. Sueur 100 772 2014 Acoustic indices for biodiversity assessment and landscape investigation. 10.3813/AAA.918757 PLoS One Sueur 3 2008 Rapid acoustic survey for biodiversity appraisal. 10.1371/journal.pone.0004065 The Calculation of Acoustic Indices Derived from Long-Duration Recordings of the Natural Environment. Towsey 2018 QutEcoacoustics/Audio-Analysis: Ecoacoustics Audio Analysis Software v20.11.2.0. Towsey 2020 10.5281/ZENODO.4274299 Proc. Comput. Sci. Towsey 29 703 2014 Visualization of long-duration acoustic recordings of the environment. 10.1016/j.procs.2014.05.063 J. Ecoacoustics Towsey 2 2018 Long-duration, false-colour spectrograms for detecting species in large audio data-sets. 10.22261/JEA.IUSWUI Landsc. Ecol. Tucker 29 745 2014 Linking ecological condition and the soundscape in fragmented Australian forests. 10.1007/s10980-014-0015-1 Ecol. Indic. Ulloa 90 346 2018 Estimating animal acoustic diversity in tropical environments using unsupervised multiresolution analysis. 10.1016/j.ecolind.2018.03.026 BMC Ecol. Ulloa 19 2019 Explosive breeding in tropical anurans: environmental triggers, community composition and acoustic structure. 10.1186/s12898-019-0243-y Landsc. Ecol. Villanueva-Rivera 26 1233 2011 A primer of acoustic analysis for landscape ecologists. 10.1007/s10980-011-9636-9 Methods Ecol. Evol. Wrege 8 1292 2017 Acoustic monitoring for conservation in tropical forests: examples from forest elephants. 10.1111/2041-210X.12730 Ecol. Inform. Znidersic 55 2020 Using visualization and machine learning methods to monitor low detectability species—The least bittern as a case study. 10.1016/j.ecoinf.2019.101014 Sci. World J. Zolhavarieh 2014 2014 A review of subsequence time series clustering. 10.1155/2014/312521