Bat echolocation plasticity in allopatry: a call for caution in acoustic identification of Pipistrellus sp.

Abstract

Animals modify their behaviours and interactions in response to changing environments. In bats, environmental adaptations are reflected in echolocation signalling that is used for navigation, foraging and communication. However, the extent and drivers of echolocation plasticity are not fully understood, hindering our identification of bat species with ultrasonic detectors, particularly for cryptic species with similar echolocation calls. We used a combination of DNA barcoding, intensive trapping, roost and emergence surveys and acoustic recording to study a widespread European cryptic species complex (Pipistrellus pipistrellus and Pipistrellus pygmaeus) to investigate whether sibling bat species could exhibit extreme echolocation plasticity in response to certain environmental conditions or behaviours. We found that P. pygmaeus occupied the acoustic niche of their absent congeneric species, producing calls with P. pipistrellus’ characteristic structure and peak frequencies and resulting in false positive acoustic records of that species. Echolocation frequency was significantly affected by the density of bats and by maternity rearing stage, with lower frequency calls emitted when there was a high density of flying bats, and by mothers while juveniles were non-volant. During roost emergence, 29% of calls had peak frequencies typical of P. pipistrellus, with calls as low as 44 kHz, lower than ever documented. We show that automatic and manual call classifiers fail to account for echolocation plasticity, misidentifying P. pygmaeus as P. pipistrellus. Our study raises a vital limitation of using only acoustic sampling in areas with high densities of a single species of a cryptic species pair, with important implications for bat monitoring. Significance statement Ultrasonic acoustic detectors are widely used in bat research to establish species inventories and monitor species activity through identification of echolocation calls, enabling new methods to study and understand this elusive understudied group of nocturnal mammals. However, echolocation call signalling in bats is intrinsically different to that of other taxa, serving a main function of navigation and foraging. This study demonstrates an extreme level of plasticity, showing large variation in call frequency and structure in different situations. We showcase the difficulty and limitation in using acoustic sampling alone for bat monitoring and the complications of setting parameters for species identification for manual and automatic call classifiers. Our observations of call frequency variation correlated with density and absence of congenerics provide novel insights of behavioural echolocation plasticity in bats.

Towards a General Approach for Bat Echolocation Detection and Classification

“…While methods for semi-supervised and self-supervised training offer the potential to learn effective models with limited to no training supervision, diverse labelled data is still needed to evaluate the performance of the developed models. Bat calls can exhibit plasticity depending on the population sampled (Montauban et al, 2021), the presence of other species, and the composition of the local environment. As a result, care needs to be taken to ensure that the collected training datasets are representative of the downstream deployment situations as much as possible.…”

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

“…For example, individuals can occupy the acoustic niche of another species by resource partitioning via call frequency displacement. Associated calls have a similar structural characteristic, which result in false‐positive acoustic records of its congeneric or conspecific species (Gager, 2019; Montauban et al, 2021). Standard occupancy models assume all false positives are removed (Mackenzie et al, 2002), and that detection from a captured bat is equal to that of a recorded call.…”

“…To best represent the variability known to be present in wild bat populations (Montauban et al, 2021; Mora et al, 2011; Obrist et al, 2004), all calls from all species were included in the classification models, despite differences in sample sizes. The call characteristics and frequencies of the Nyctophilus species in this region almost completely overlap (Adams et al, 2010) and it is therefore not suitable to identify species within this genus from acoustic data alone.…”

Wind energy development can lead to guild‐specific habitat loss in boreal forest bats

“…In 2021, Pipistrellus pipistrellus was designated as 'Not Applicable' on the Norwegian Red List (Eldegard et al 2021) and thus this species was reclassified as unknown pipistrelle species due to similarities in call structure with other Pipistrellus spp. (Montauban et al 2021).…”