Bats are still not birds in the digital era: echolocation call variation and why it matters for bat species identification

Abstract

The recording and analysis of echolocation calls are fundamental methods used to study bat distribution, ecology, and behavior. However, the goal of identifying bats in flight from their echolocation calls is not always possible. Unlike bird songs, bat calls show large variation that often makes identification challenging. The problem has not been fully overcome by modern digital-based hardware and software for bat call recording and analysis. Besides providing fundamental insights into bat physiology, ecology, and behavior, a better understanding of call variation is therefore crucial to best recognize limits and perspectives of call classification. We provide a comprehensive overview of sources of interspecific and intraspecific echolocation call variations, illustrating its adaptive significance and highlighting gaps in knowledge. We remark that further research is needed to better comprehend call variation and control for it more effectively in sound analysis. Despite the state-of-art technology in this field, combining acoustic surveys with capture and roost search, as well as limiting identification to species with distinctive calls, still represent the safest way of conducting bat surveys.

Social communication in bats

“…Bats are known to use two types of sounds for communication, namely echolocation and social calls (Fenton, ; Dechmann & Safi, ; Gillam & Fenton, ; Russo et al, , Fig. ).…”

Improving geographically extensive acoustic survey designs for modeling species occurrence with imperfect detection and misidentification

“…It is also debatable whether hand captures constitute true detections for certain bat species, as many species are morphologically cryptic (e.g., Rodhouse, Scott, Ormsbee, & Zinck, ; Rodriguez & Ammerman, ; Weller, Scott, Rodhouse, Ormsbee, & Zinck, ). Bat acoustic data pose challenges for the calibration model as well, as the libraries of echolocation calls used to build classification software for automated species identifications are not always made under realistic conditions (Russo et al., ). The OC models, on the other hand, show promise for leveraging information from bat acoustic surveys while potentially reducing the manual verification burden.…”

“…Broad-scale monitoring programs have been initiated across Europe, North America, and elsewhere (e.g., Barlow et al, 2015;Jones et al, 2013;Loeb et al, 2015;Roche et al, 2011;Walters et al, 2012) and rely in part on coordinated acoustic surveys. The complication is that shared echolocation call characteristics from morphologically and ecologically similar bat species can result in incorrect species assignments from automated identification software and misidentification errors for the focal species (Russo, Ancillotto, & Jones, 2017;Russo & Voigt, 2016;Rydell, Nyman, Eklof, Jones, & Russo, 2017). Imperfect detection can occur when all echolocation calls from the focal species are of such low quality that they are filtered out during call processing.…”

Scientific statement on the coverage of bats by the current pesticide risk assessment for birds and mammals

“…Echolocation call design is tailored to exploit different habitat structures, which explains why it differs among species (e.g. Jones and Holderied ; Russo et al., ,b,c). The species that hunt in open space broadcast echolocation calls at lower frequencies, and exhibit narrower frequency bandwidths, than those hunting in cluttered habitat.…”