This dataset belongs to Dr. Forest's task T1.2 "Sensory experience of biodiversity via soundscapes." and contains mean values for daytime (dawn till dusk) acoustic indices in spring 2021 and 2022. Its is derived from the dataset: "Dr.Forest T1.2 Acoustic Indices from audio recordings_complete dataset" and will be used in context of the BBNSs. Bird richness was assess by an expert that manually listened to a subset of the audiofiles and identified bird species. For Dr.Forest study sites Bialowieza, Hainich, TreeWeb, Forbio (Gedinne, Zegeldem, Hechtel-Eksel) and Biotree (Kaltenborn) data is available for spring 2021, for the sites Bialowieza, Hainich, TreeWeb, MixLor data is available for spring 2022, Bird richness was assessed only in one year.

see Sampling protocoll T1.2 for the set up and recording specifications.

Poland, Bialowieza: spring 2021: 15.04.2021-15.05.2021; spring 2022: 15.04.2022-15.05.2022// Germany, Hainich: spring 2021: 06.04.2021-06.05.2021; spring 2022: 15.04.2022-15.05.2022// Belgium, TreeWeb: spring 2021: 15.04.2021-15.05.2021; spring 2022: 14.04.2022-14.05.2022// France, MixLor, four species platform: spring 2022: 13.04.2022-05.05.2022 (stopped early because if SD card issues)// Forbio, Zegedem: spring 2021: 15.04.2021-15.05.2021// Forbio, Hechtel-Eksel: spring 2021: 15.04.2021-15.05.2021// Forbio, Gedinne: spring 2021: 24.04.2021-24.05.2021// Biotree, Kaltenborn: spring 2021: 20.04.2021-20.05.2021//

Birds && Acoustic Indices, these are based on acoustic activity of the abiotic and biotic environment. In this specific case, most acoustic events are either from wind or bird vocalisation. We derived BirdBiophony, Silence and OTHER (which was mainly Geophony, in form of wind and rain) as measures to seperate the dirfferent soundscape components (see Analysis for details).

Acoustic indices were calculated using the "Analysis Program" (Towsey et al., 2018). "High Amplitude" and "Clipping Index" were used to identify recordings with very loud wind and rain sounds and exclude all recordings with values >0 from this dataset. The indices were calculated based on a 1-minute recording length. The two consecutive minutes per recording were averaged, resulting in one index value per 15-minute period. For a detailed description of the acoustic indices and how they are calculated in the analysis program, see Towsey et al. (2018): Towsey, M., Truskinger, A., Cottman-Fields, M., and Roe, P. (2018). Ecoacoustics audio analysis software. Version V18.03.0.41. Zenodo. https://doi.org/10.5281/zenodo.1188744. In addition, we specified the percentage of audio recordings that contained bird sounds (BirdBiophonypercent) based on BirdNET specifications (https://ravensoundsoftware.com/knowledge-base/birdnet-analyzer/) using default settings, and a virtual species list based on site coordinates and time of recording (middle week of recording period, which was week 17 of the year. Silence was indicated when two indices were below a certain tresholde (EventsPerSecond < 0.02501 & EntropyOfCoVSpectrum < 0.025 -> "silence")). The remaining recordings were labeled "OTHER" and contained mainly wind and rain (geophony),which was confirmed by manual aural inspection. The manual aural inspection of a random subsample also revealed that almost all wave files that contained bird sounds were correctly labeled as "BirdBiophony", but sometimes they additionaly contained wind or rain sounds. Also, most of the files labeled "silence" contained no discernible sounds, and only sometimes contained very soft bird calls or wind noise that was barely distinguishable from background noise.