A machine learning framework to classify Southeast Asian echolocating bats

Yoh, Natalie; Kingston, Tigga; McArthur, E. Durant; Aylen, Oliver E.; Huang, Joe Chun‐Chia; Jinggong, Emy Ritta; Khan, Faisal Ali Anwarali; Lee, Benjamin P. Y.‐H.; Mitchell, Simon L.; Bicknell, Jake E.; Struebig, Matthew J.

doi:10.1016/j.ecolind.2022.108696

Cited by 12 publications

(14 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The results from the cross‐validation reported near‐perfect specificity across all species (>0.9), and this can be especially problematic for elusive or harder to identify species (Rojas et al, 2019). Metrics such as specificity highlight the certainty of true negatives, and this can be unreliable in multicategory classifications (López‐Baucells et al, 2019; Yoh et al, 2022).…”

Section: Discussionmentioning

confidence: 99%

“…Secondly, we measured classification success for each species by comparing the manual identification (Dataset 4), against the automatic identification assigned by each classifier. We hypothesized that manual identification provided the most conservative species identifications, which we used to assign true (correct automatic identification) or false positive (failed automated identification) to the final predictions (Barré et al, 2019; López‐Baucells et al, 2019; Yoh et al, 2022). From these, we could predict the confidence score corresponding to a given success probability of the automated identification.…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

A new method employing species‐specific thresholding identifies acoustically overlapping bats

2022

View full text Add to dashboard Cite

Passive acoustic detectors are increasingly used for monitoring biodiversity, particularly for echolocating bat species (Microchiroptera). However, identification of calls collected at large scales is hindered by substantial variation within and between species, and the considerable time investment needed to manually identify acoustic data. We use acoustic data from 14 species of echolocating bats, occurring in temperate forests and woodlands of southeastern Australia to build a supervised classification model that identifies species from large acoustic datasets. Acoustic data from hand‐release (39,567) and free‐flying (8851) bat calls were used to build a predictive model, which was then validated using field‐collected calls (149,097) from the same region. We maximized the model fit per species by validating the associated confidence scores against manually identified presence and absence values. This allowed us to model the identification success of each species as a function of the confidence score. From this relationship, we set specific thresholds for accepting species identification, enabling more accurate classification of calls and identification of multiple bat species within a single acoustic recording. Including calls from manually identified free‐flying bats improved overall identification accuracy, including a 60% improvement for bats that navigate in open spaces. Assigning species‐specific thresholds achieved substantial improvements in overall model confidence, with functionally meaningful changes in the identification of species exhibiting considerable acoustic overlap in time and frequency measures. Research into the ecological requirements of species is hampered by problems with identification. Our research illustrates that internal train–test validation overestimates model accuracy particularly for species that were in low abundance or for uncommon species, which are acoustically similar to more common ones. Recognizing this, we set specific thresholds per species below which identifications were not accepted. Our method is particularly relevant in locations with high overlap in species' call parameters, which can result in false negatives in preference for species that are easier to identify because of the common practice of assigning one species per acoustic recording. This research proposes a cautious method to substantially reduce the burden of manual identification of large acoustic datasets.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

A new method employing species‐specific thresholding identifies acoustically overlapping bats

2022

View full text Add to dashboard Cite

show abstract

“…For those species we could not find any reference calls, we used the reference calls available in the bat call library Chirovox (Görföl et al, 2022). We therefore matched the call type of the species present in peninsular Malaysia to one of the sonotypes described in Yoh et al (2022) for bat species in Malaysian Borneo, namely CF, FM, FMqCF1, FMqCF2, FMqCF3, FMqCF4, FMqCF5, QCF and LF sonotypes. Using start and end frequency, frequency of maximum energy, duration and interpulse interval as defined in Russo and Jones (2002), we classified the calls into one of these nine sonotypes.…”

Section: Methodsmentioning

confidence: 99%

“…As the shape of the echolocation calls reflects the physical constraints encountered by the bats, we were able to classify bat sonotypes into three foraging guilds: (1) the constant-frequency and FM calls represent forest foragers, (2) FMqCF4, FMqCF5 and QCF represent edge foragers, and (3) LF, FMqCF2 and FMqCF3 calls represent open-space foragers (Yoh et al, 2022). Social calls could not be identified to the sonotype level and were treated as assemblage-level activity.…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Foraging guild modulates insectivorous bat responses to habitat loss and insular fragmentation in peninsular Malaysia

Hazard

Froidevaux

Yoh

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

Despite mounting evidence on the ecological impacts of damming for biodiversity, little is known regarding its consequences in the hyper-diverse Southeast Asian tropical forests. Here we assess the effects of habitat loss and fragmentation on the diversity and activity of insectivorous bats within the hydroelectric Kenyir Lake in peninsular Malaysia. We surveyed bat assemblages on 26 islands and two mainland continuous forest sites using passive acoustic monitoring. Echolocation calls were classified into sonotypes, each corresponding to either one or multiple species, and grouped into foraging guilds. We then examined bat overall assemblage (sonotype richness, activity, and composition), guild- and sonotype-specific activity. From 9360 hours of recordings, we identified 16 bat sonotypes, including 10 forest (2854 bat passes), three edge (13 703) and three open-space foragers (3651). Sonotype richness increased towards denser forest structures (higher Normalized Difference Vegetation Index - NDVI), while species composition varied across the gradient of forest area. Forest foragers were positively affected by NDVI and negatively affected by distance to the closest neighbour, whereas edge foragers' activity increased in smaller islands. Of the six sonotypes analysed, the activity of one forest sonotype increased with forest area, while that of one edge sonotype decreased. Ensuring habitat quality within insular forest remnants, in addition to their functional connectivity, maximises bat diversity, including the persistence of forest foraging species. Future hydropower development should therefore avoid the creation of a myriad of small, isolated, and habitat-degraded islands further characterised by altered levels of bat diversity and guild-level activity.

show abstract

Foraging guild modulates insectivorous bat responses to habitat loss and insular fragmentation in peninsular Malaysia

Hazard

Froidevaux

Yoh

et al. 2023

Biological Conservation

View full text Add to dashboard Cite

A machine learning framework to classify Southeast Asian echolocating bats

Cited by 12 publications

References 48 publications

A new method employing species‐specific thresholding identifies acoustically overlapping bats

A new method employing species‐specific thresholding identifies acoustically overlapping bats

Foraging guild modulates insectivorous bat responses to habitat loss and insular fragmentation in peninsular Malaysia

Foraging guild modulates insectivorous bat responses to habitat loss and insular fragmentation in peninsular Malaysia

Contact Info

Product

Resources

About