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

Bird biodiversity estimation within a fixed detection radius The effectiveness of bird species detection by a recorder in comparison with human-observer surveying birds within a fixed radius has been examined previously [30,32]. In such approach the distance to a bird is estimated by the observer, while the recorder detects birds without any distance limitation. Our study showed that after applying a fixed detection radius by the human-observer (50 m or 100 m) the recorder detected significantly more bird species than the observer. It means that the effective range of detection by a recorder for most bird species exceeds 100 m (Table 1, Figs 1 and S1). Looking at a single species (Table 2, S1 Fig) we can see that detection rate for most of them is higher by recorder than by human-observer surveying birds within 50 m and 100 m distance (36 and 23 species, respectively), while in unlimited dis- tance the most species are detected similarly by observer and recorder. The visual and audio detection rate of a species depends on their behaviour and call charac- teristics. In general, species that are small, hidden or difficult to visually recognise (e.g., Com- mon whitethroat Curruca communis, Common quail Coturnix coturnix, Marsh warbler Acrocephalus palustris) are detected similarly by recorder and observer from 100 m or unlim- ited distance, dependently on the loudness of their vocalisation, while detection rate of easily seen species including small (e.g., Whinchat Saxicola rubetra) and large ones (e.g, White stork

Ciconia ciconia or Common crane Grus gus), is higher for human-observers within unlimited distance than for recorders. However, it is possible to find species (Table 2) or groups of spe- cies (e.g., meadow or farmland species; Table 1, Fig 1) that are detected with a similar detection rate within a 100 m radius both for the recorder and human-observer. Furthermore, the overall bird species richness is not necessarily the best indicator when monitoring is conducted in a very specific environment [43]. Therefore, we examined the effectiveness of acoustic recorders in the estimation of meadow and farmland bird biodiversity–two groups of birds specific to meadow habitat and important from the perspective of monitoring of agricultural landscape. In both groups we found that the recorder and human-observer surveying birds within a 100 m radius provide similar estimations of bird biodiversity. Therefore, the traditional observer- based monitoring of these two groups of birds can be successfully substituted with acoustic monitoring at least in this regional habitat, even without any corrections for acoustic proper- ties of environment, habitat type and detection distance of a particular species. Applying acoustic monitoring of meadow species in practice Our results allow to propose acoustic monitoring as equally accurate and potentially more effi- cient alternative to traditional field surveys conducted by ornithologists in agricultural land- scapes. The first advantage of acoustic monitoring is that data can be collected during a short

breeding season in the field by non-specialists (e.g., local agriculture management authorities, landowners) in very standardized way (time during a season and day, survey duration, weather conditions, equipment), and analysed later in the lab. Therefore, acoustic monitoring enables us to sample incomparably more locations than by using traditional observer-based approach. The second advantage is standardised acoustic data analysis. In our approach we applied the simplest method–manual spectrogram scanning and listening to recordings. Even using this method, collected during the short breeding season data can be analysed in the rest part of the year. However, applying acoustic monitoring at regional or national level enables the use of automatic species detection and recognition algorithms [50,51] or applying acoustic indicators [52], which is more efficient and more acceptable for incorporation into big data analysis. Moreover, soundscape recordings can be used to monitor other vocalising taxa, such as insects or amphibians [7,53], whose importance is increasingly appreciated but the monitoring is still insufficient and difficult to conduct. In the simplest way acoustic monitoring could be, for PLOS ONE Acoustic bird biodiversity monitoring in agricultural meadows PLOS ONE | https://doi.org/10.1371/journal.pone.0266557 April 8, 2022 11 / 16