Testing the performances of automated identification of bat echolocation calls: A request for prudence

Rydell, Jens; Nyman, Stefan; Eklöf, Johan; Jones, Gareth; Russo, Danilo

doi:10.1016/j.ecolind.2017.03.023

Cited by 108 publications

(98 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…semi‐automated identification; Newson et al., ), is a key tool for improving reliability of studies based on acoustic data. Indeed, robust ecological responses could be produced even in cases where false positive rates were so far considered too high (Rydell et al., ). This new and robust framework takes advantage of confidence scores provided by the automated identification software and its ability for distinguishing true positives and false positives (Figure S3), controlling for false positive tolerances (FPTs), and checking for potential biases induced by identification errors.…”

Section: Discussionmentioning

confidence: 99%

“…There is thus an implicit relationship between the error rate and confidence scores and most software manuals advocate using confidence thresholds below which data should be discarded to minimize the error rate, for example Tadarida (Bas et al., ), SonoChiro (Biotope, ) and BatClassify (Scott & Altringham, ). Regardless of the software used, the relationship between the error rate and confidence scores is an important part of the automated identification performance, yet it has never been directly assessed in previous methodological studies (Fritsch & Bruckner, ; Rydell et al., ). Consequently, the level at which confidence thresholds should be set is unclear to most users, which has limited the use of automated identification in ecological studies.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Accounting for automated identification errors in acoustic surveys

et al. 2019

View full text Add to dashboard Cite

Assessing the state and trend of biodiversity in the face of anthropogenic threats requires large‐scale and long‐time monitoring, for which new recording methods offer interesting possibilities. Reduced costs and a huge increase in storage capacity of acoustic recorders have resulted in an exponential use of passive acoustic monitoring (PAM) on a wide range of animal groups in recent years. PAM has led to a rapid growth in the quantity of acoustic data, making manual identification increasingly time‐consuming. Therefore, software detecting sound events, extracting numerous features and automatically identifying species have been developed. However, automated identification generates identification errors, which could influence analyses which look at the ecological response of species. Taking the case of bats for which PAM constitutes an efficient tool, we propose a cautious method to account for errors in acoustic identifications of any taxa without excessive manual checking of recordings. We propose to check a representative sample of the outputs of a software commonly used in acoustic surveys (Tadarida), to model the identification success probability of 10 species and two species groups as a function of the confidence score provided for each automated identification. Using this relationship, we then investigated the effect of setting different false positive tolerances (FPTs), from a 50% to 10% false positive rate, above which data are discarded, by repeating a large‐scale analysis of bat response to environmental variables and checking for consistency in the results. Considering estimates, standard errors and significance of species response to environmental variables, the main changes occurred between the naive (i.e. raw data) and robust analyses (i.e. using FPTs). Responses were highly stable between FPTs. We conclude it was essential to, at least, remove data above 50% FPT to minimize false positives. We recommend systematically checking the consistency of responses for at least two contrasting FPTs (e.g. 50% and 10%), in order to ensure robustness, and only going on to conclusive interpretation when these are consistent. This study provides a huge saving of time for manual checking, which will facilitate the improvement in large‐scale monitoring, and ultimately our understanding of ecological responses.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Accounting for automated identification errors in acoustic surveys

et al. 2019

View full text Add to dashboard Cite

show abstract

“…In many cases, acoustic systems have been developed for automated species classification of huge volumes of call data. In the case of bats, echolocation calls are not songs, making the identification to species from bat calls a challenging exercise that requires suitable cross-testing of results using reliably identified calls, e.g., from captured and released individuals (Barclay 1999; Taylor et al 2013; Monadjem et al 2017; Rydell et al 2017). Similarly, camera traps have enabled efficient and comprehensive surveys of medium and large-sized mammals and other groups (Stein et al 2008; Tobler et al 2008; Rovero and Marshall 2009; Rovero et al 2014).…”

Section: Introductionmentioning

confidence: 99%

Tapping into technology and the biodiversity informatics revolution: updated terrestrial mammal list of Angola, with new records from the Okavango Basin

Taylor

Neef²,

Keith³

et al. 2018

View full text Add to dashboard Cite

Using various sources, including the Global Biodiversity Information Facility (GBIF), published literature, recent (2015–2017) collections, as well as bat detector and camera trap surveys with opportunistic sightings and live capture in the upper Okavango catchment in central Angola, we present an updated mammal checklist of 275 species from 15 different orders for Angola (including the Cabinda region). Recent surveys (captures and bat detectors) of small mammals from the upper Okavango catchment yielded 46 species (33 species of bats, ten species of rodents and three species of shrews). One bat (Pipistrellusrusticus, rusty pipistrelle); two rodents (Mussetzeri, Setzer’s mouse and Zelotomyswoosnami, Woosnam’s broad-faced mouse) and one shrew (Suncusvarilla, lesser dwarf shrew) were captured for the first time, in Angola. While our species lists of bats conformed to predicted totals, terrestrial small mammals were under sampled, with only 13 species recorded by our trapping survey compared to a total of 42 shrew and rodent species expected based on GBIF records for the central Angolan highlands. Seven terrestrial small mammal species (one shrew and six rodents) are endemic to the central and western Angolan highlands but none of these were captured in our survey. The bat detector surveys added three further bat species to the country list: Pipistrellushesperidus, Kerivoulaargentata, and Mopsmidas. Camera trap surveys and opportunistic sightings in the upper Okavango catchment in 2016 yielded a total of 35 species of medium-large mammals, from 17 families, although all of these had been reported previously in Angola. GBIF proved to be an excellent source of biodiversity data for Angolan mammals, most importantly for documenting dramatic historical range changes of larger mammals such as the sable (Hippotragusnigerniger), Kirk’s sable (H.nigerkirkii) and the giant sable (H.nigervariani).

show abstract

“…To classify bat activity (i.e., passes) to family and genus post file-conversion and scrubbing, we used two methods: (1) automated identification using Kaleidoscope Pro 3 software and associated Neotropical and North American bat classifiers; and (2) visual identification of spectrograms [42][43][44][45][46][47][48] displayed in the Kaleidoscope Pro 3 Viewer. When classifying bat passes using automated and visual identification, differentiating among species' calls can be difficult; it is affected by the degree of clutter at sampling locations, direction the bat is pointing relative to the microphone when it emits a call, angle and direction of the detector microphone, call attenuation, Doppler shift, and similarity of call characteristics of different species [49][50][51][52]. Therefore, to avoid misidentification, we only classified recorded bat passes to family and genus.…”

Section: Bat Monitoringmentioning

confidence: 99%

Forest Structure and Composition Affect Bats in a Tropical Evergreen Broadleaf Forest

Willcox

Giuliano

Watine

et al. 2017

Forests

View full text Add to dashboard Cite

Abstract:The lack of knowledge regarding many aerial insectivorous bats and their relationships with forest characteristics limits conservation decision-making for tropical rainforests and for this important bat group. Therefore, our objective was to understand the effects of forest structure and composition on these bats in the Neotropical evergreen broadleaf forest of Belize, Central America. We conducted bat monitoring and quantified 51 forest characteristics at 24 locations in the Chiquibul Forest Reserve (CFR) from May-July 2014. Simple linear and backward stepwise multiple regression analyses were used to examine relationships between bat richness and activity and forest characteristics. Bat genus richness and total activity were directly related to overstory canopy depth and inversely related to ≤4 structural characteristics. Lasiurus, Myotis, Promops, and Pteronotus spp. were affected by ≤7 forest characteristics; the responses were explained by preferences for less-cluttered, open space for flying and foraging and species-specific food and cover requirements. However, bat richness and activity were often unaffected by forest structure and composition in the CFR, suggesting that at this taxonomic level, bats may not be very sensitive to variation in forest characteristics, may not be very useful indicators of alteration, and may have some tolerance for disturbance and change.

show abstract

Testing the performances of automated identification of bat echolocation calls: A request for prudence

Abstract: General rightsThis document is made available in accordance with publisher policies. Please cite only the published version using the reference above. Full terms of use are available: http://www.bristol.ac.uk/pure/about/ebr-terms

Cited by 108 publications

References 19 publications

Accounting for automated identification errors in acoustic surveys

Accounting for automated identification errors in acoustic surveys

Tapping into technology and the biodiversity informatics revolution: updated terrestrial mammal list of Angola, with new records from the Okavango Basin

Forest Structure and Composition Affect Bats in a Tropical Evergreen Broadleaf Forest

Contact Info

Product

Resources

About