Acoustic approach as an alternative to human-based survey in bird biodiversity monitoring in agricultural meadows

Introduction Birds are useful biological indicators of environmental conditions and their changes [1,2]. There- fore, conservation biologists and ecologists monitor bird populations in various environments, from a regional to a global spatial scale [3,4]. In the traditional human-based approach, highly skilled observers survey birds in the field by detecting and recognising them aurally and visually [5]. The most restrictive monitoring schemes require multiple surveys of study area done by observers with many years of experience in order to detect all territories and minimise the observ- er’s effect on survey results [6]. However, the number of highly skilled observers in a short breed- ing season is limited, and so this type of monitoring is usually conducted on a small, local scale. The large-scale monitoring schemes such as regional or national often include less experienced observers who possess varied skill sets and usually survey the study area twice in the breeding sea- son, which entails a compromise between data quality and the spatial extent of monitoring. Acoustic monitoring has been proposed as an effective alternative for traditional, observer- based approach [7–12]. The acoustic monitoring process combines data collection in the field with laboratory analyses of soundscape recordings [10,13,14]. An advantage of acoustic moni- toring is that data can be collected in a highly standardised way in many locations at the same time [10], without highly skilled fieldworkers, observer effect, mistakes made in terms of spe-

cies detection and identification in the field [15,16]. The acoustic monitoring is also low-cost in comparison to traditional observed-based approaches and allows for the generation of more data [10,17]. The final advantage is that original data are collected and stored, enabling the ver- ification of correctness of species detections or for the re-analysis of recordings in the future using new detection methods and techniques [10]. Due to the possibility of repeated listening or spectrogram scanning, there is less chance to overlook the species or make an identification error [18], which is possible in traditional, human-based field surveys, especially when many species vocalise at the same time and often are observed for a short time [19,20]. Acoustic monitoring also has a few limitations, which should be considered. The first is that some bird species vocalise rarely or are completely silent during some parts of a day or season, thus cannot be detected by a recorder, while a human-observer is able to detect them visually [19,21]. However, this limitation can be easily broken by increasing the duration of the recording time or appropriate sampling across a day and season [22,23]. The second is the diversity of equipment used for soundscape recording [10,24]. Using recorders and micro- phones of different quality leads to the same sound recorded in the same habitat and weather conditions being recorded from a different distance by various recorders, thus sampling area is inconstant [24]. Another issue is the variation in amplitude and frequency of sounds pro-

duced by birds. Birds, depending on the species, produce sounds in wide range of frequency, from several dozen Hz to even 10 kHz [25]. Low-frequency sounds are generally transmitted for longer distance than high-frequency ones, thus considerable between-species variation in detection range is observed [24]. Independently of the frequency of sound, the loudness of birds’ vocalisation ranges from ca 50 to even 120 dB measured at 1 m [26,27]. These two char- acteristics of different bird species vocalisations result in the detection range being species-spe- cific [28]. Moreover, the detection range strongly depends on habitat type—probability of species detection at a given distance is higher in open vegetation than in forest [29]. Therefore, it is important to know how effective the acoustic technique is in a specific environment inhab- ited by a composition of bird species which are characterised by various probabilities of acous- tic detection and various distances from which the recorder can detect their vocalisations. Comparisons of species detection by observer and recorder have been conducted around the world, mostly in forest or semi-forest habitats [19,21,23,30,31]. In most cases, researchers compared bird species richness from the same point and at the same time by both acoustic PLOS ONE Acoustic bird biodiversity monitoring in agricultural meadows PLOS ONE | https://doi.org/10.1371/journal.pone.0266557 April 8, 2022 2 / 16