Data and code from: Forest degradation reduces acoustic partitioning in tropical avian communities
Data files
Jun 14, 2026 version files 2.54 MB
-
ANH_code_upload.R
30.71 KB
-
Ecoacoustic_sample_results_to_share_2025.csv
1.85 MB
-
README.md
3.12 KB
-
selections6.csv
647.20 KB
-
transect_gradients.csv
899 B
Abstract
The acoustic niche hypothesis posits that species partition acoustic space by frequency and time to avoid signal masking, and that any local loss of species leaves detectable gaps in co-evolved soundscapes. However, loss, fragmentation and disturbance of habitat often lead to species turnover, and as sensitive species are lost, generalist species colonise, potentially leading to novel overlap of acoustic niches. We tested the acoustic niche hypothesis in varyingly disturbed tropical forests in the Brazilian Amazon by comparing overlapping vocalisations in undisturbed and disturbed forest. These datasets contain overlapping calls per 15-second sound file (selections6.csv) and species labelled per 15-seond sound file (Ecoacoustic_sample_results_to_share_2025.csv) which were used in analysis where we measured the effect of disturbance on the frequency of overlapping vocalisations, then assigned sensitivity scores to each species and modelled how species sensitivity and species richness per sound file impacted overlapping vocalisations.
Dataset DOI: 10.5061/dryad.pvmcvdp1q
Description of the data and file structure
Ecoacoustic sample results to share are the species names identified in each sound file (matching those in selections6). The 'method' column relates to another study on the efficacy of high vs low temporal-resolution sampling. During the analysis, low temporal-resolution files are removed; otherwise, this is not relevant to this study.
transectgradients.csv contains information about the type and age of disturbance each transect has been exposed to.
Files and variables
File: ANH_code_upload.R
Description: Associated R script for analysis
ANH_code_upload is the R script that contains all analyses from the study. Input of all the required files is already included in the script, and there are annotations throughout to explain which part of the analysis is handled in each section, starting with the NMDS for assigning species sensitivity score, modelling of the number of overlapping vocalisations between forest classes, and finally modelling how species richness and sensitivity score change with the number of overlapping vocalisations.
Input files for this code are:
selections6.csv contains several overlapping vocalisations from manual labelling of 15-second sound files collected using passive acoustic monitoring in the Brazilian Amazon. Columns include the unique transect name, disturbance type (as described in the accompanying paper), date and time of collection, number of overlaps in that file (selections), and species richness of birds in that file (sr).
File: Ecoacoustic_sample_results_to_share_2025.csv
Description:
Variables
- Transect: unique transect ID
- Method: high or low temporal resolution sampling. Low temporal resolution was removed during analysis. Not relevant to this study.
- Date: date of sound file recording
- Time: time of sound file recording
- Sp: species name
File: selections6.csv
Description:
Variables
- transect: unique transect ID (matching those above)
- disturbance_type: type of anthropogenic disturbance transect has been exposed to (as described in the study)
- file_no: arbitrary numbering of files
- date: date sound file collected
- time: time sound file collected
- selections: number of overlapping vocalisations in that sound file
- sr: species richness - number of vocalising avian species identified in that sound file
- prop: from previous analysis, irrelevant for this study
- survey: combined transect and date/time for unique survey ID
File: transect_gradients.csv
Description:
Variables
- Transect: unique transect ID
- Gradient: descriptive title of disturbance type
- Logged: true or false - transect previously logged?
- Secondary: true or false - transect a secondary forest?
- Burned: true or false - transect previously burned?
- Age of Burn: New = recent burn (2015), Old = old burn (prior to 2015), Both = 2015 and prior