https://doi.org/10.5061/dryad.p2ngf1vzw

The data and code presented in this repository supports the analysis in Luypaert et al. (2024): Extending species-area relationships to the realm of acoustics - the soundscape-area relationship. The repository contains all required data and R scripts to calculate the soundscape richness acoustic index (alpha, beta and gamma), and assess the spatial variation of this index in relation to island size and isolation at the Balbina Hydroelectric Reservoir in Brazilian Amazonia. 

Usage note: Download the 'Luypaert_et_al_2024_RNotebooks.zip'  file from this repository and unzip it in the desired location. If you are only interested in inspecting the code and outputs without running it, just inspect the '.html' files in sequential order from 1 to 10. If you want to try running the code, simply start with RNotebook 1 (1_Metadata_preparation.Rmd) and follow the order of the listed RNotebooks (1 to 10) to create all plots and models presented in the main text and supplementary materials of Luypaert et al. (2024). All RNotebooks are self-contained and automatically locate the required data objects and scripts for analysis. 

Note on downloading the data: Some users may experience difficulties downloading or unzipping the data in this repository. If you are experiencing issues with the data download, be patient and only click the download button once to avoid overwhelming the Dryad download prompt. This dataset contains many thousands of files, corresponding to the individual CVR-index value files for each 1-minute file in the study, which can lead to longer download times. If you experience difficulties unzipping the downloaded data file, particularly on Mac computers, consider using the following data unpacking software. If you still experience issues after following these steps, you can reach out at thomas.luypaert@nmbu.no.

Description of the data and file structure

1. File structure

Below, we describe the file structure for 'Luypaert_et_al_2024_RNotebooks.zip', available for download from this repository. Filenames are listed in italic, whereas directory names are listed in bold. The nested hierarchy is represented by the order of  bullet points. Where too many files are present in a directory, the content is briefly described. 

Luypaert_et_al_2024_RNotebooks.zip

• "1_Metadata_preparation.Rmd"
• "1_Metadata_preparation.html"
• "2_Soundscape_metrics.Rmd"
• "2_Soundscape_metrics.html"
• "3_Map.Rmd"
• "3_Map.html"
• "4_Small_island_effect.Rmd"
• "4_Small_island_effect.html"
• "5_Isolation_scale_of_effect.Rmd"
• "5_Isolation_scale_of_effect.html"
• "6_Data_exploration.Rmd"
• "6_Data_exploration.html"
• "7_Modelling_effect_of_area_and_isolation.Rmd"
• "7_Modelling_effect_of_area_and_isolation.html"
• "8_Decomposing_SSARs_with_rarefaction.Rmd"
• "8_Decomposing_SSARs_with_rarefaction.html"
• "9_Rarefaction_sensitivity_analysis.Rmd"
• "9_Rarefaction_sensitivity_analysis.html"
• "10_SAC_analysis.Rmd"
• "10_SAC_analysis.html"
• Data
  • Index_files
  • 69 directories (in per plot) containing 1,440 CVR index files ('.csv' format)
  • Metadata
  • "metadata_sites.csv"
  • "metadata_clean.csv"
  • "metadata_isolation.csv"
  • QGIS_data
  • Shapefiles
    • "amazonia.shp"
    • "Balbina_all_islands_without_mainland.shp"
    • "Balbina_islands_without_study_island_or_mainland.shp"
    • "Balbina_mainland_without_islands.shp"
    • "Balbina_study_island_sampling_points.shp"
    • "Balbina_study_islands.shp"
    • "islands_archipelago.shp"
    • "reservoir_balbina.shp"
  • Total_buffer_area
    • 40 '.csv' files for isolation variables (PW; 50 - 2000m buffers)
  • Clipped_buffer_area
    • 40 '.csv' files for isolation variables without focal island (PW; 50 - 2000m buffers)
  • Functions
  • A selection of additional functions that are sourced in the RMarkdown documents listed above. 
  • RData
  • "balbina_soundscapes.rds"
  • "sample_list.rds"
  • "inc_mat_island.rds"
  • "inc_mat.rds"
  • "unrarefied_gamma_soundrich.rds"
  • "rarefaction_cmin.rds"
  • "rarefaction_cmax.rds"
  • "rarefaction_cmin_total.rds"
  • "rarefaction_cmax_total.rds"

2. File descriptions

2.1. Scripts

To perform the analysis yourself, the only files that need  to be opened are those ending in '.Rmd', representing the RNotebooks containing the code for different steps in the analytical pipeline. Note that these have also been knitted into readable '.html' documents for easier inspection. The following scripts are included in the repository:

• 1_Metadata_preparation.Rmd Used to prepare all metadata for subquent analyses, including the area and isolation predictor variables. 
• 2_Soundscape_metrics.Rmd Used to calculate the 'soundscape richness' acoustic index and incidence matrices used in downstream analysis.
• 3_Map.Rmd Used to create Figure 1, containing a map of the study area and sampling plots. 
• 4_Small_island_effect.Rmd Used to test for the presence of a small island effect. 
• 5_Isolation_scale_of_effect.Rmd Used to assess at which spatial scale (50-2000m) isolation measured as the proportion of water in the surrounding buffer best describes the variation in the unrarefied gamma soundscape richness. 
• 6_Data_exploration.Rmd Used to assess variables distributions and correlations, and explore the influence of island size and isolation on the unrarefied gamma soundscape richness using partial regression plots and conditioning plots. 
• 7_Modelling_effect_of_area_and_isolation.Rmd Used to model the effect of island size and isolation on the unrarefied gamma soundscape richness. 
• 8_Decomposing_SSARs_with_rarefaction.Rmd Used to calculate the rarefied gamma,  alpha, and beta soundscape metrics and assess how they're influenced by island size. 
• 9_Rarefaction_sensitivity_analysis.Rmd Used to assess the sensitivity of SSAR model parameter estimates to the width of rarefaction confidence intervals. 
• 10_SAC_analysis.Rmd Used to check for the presence of spatial autocorrelation in the data. 

2.2. Data overview and relationships

In addition to the RNotebooks, the repository also contains a number of data objects that are called in the RNotebooks, or created by the RNotebooks. In theory, you do not need to access these files separately, as they are automatically detected and used by the scripts presented above. 

• Index_files: This directory contains 69 subdirectories - one for each plot in the study. Each subdirectory contains 1,440 '.csv' files holding the CVR index values for each 1-minute sound file that was recorded at that location. These CVR index files are used in RNotebook 2 '2_Soundscape_metrics.Rmd' to calculate the unrarefied soundscape richness and OSU-by-sample incidence matrices used in downstream analyses. 
• Metadata:
  • metadata_sites.csv A '.csv' file containing some important metadata of the sampling campaign, including the geographic coordinates of sampling plots, first day of recording per plot, number of recording days per plot, island size and island/plot names. This data object is used in RNotebook 1 '1_Metadata_preparation.Rmd' .
  • metadata_clean.csv A '.csv' file created in RNotebook 1 '1_Metadata_preparation.Rmd' . Contains metadata for the study sites used in Luypaert et al. (2024). 
  • metadata_isolation.csv A '.csv' file created in RNotebook 1 '1_Metadata_preparation.Rmd' . Contains metadata for the different isolation metrics used in downstream analyses for each of the 40 buffer distances around each island. 
• RData:
  • balbina_soundscapes.rds A list of soundscape objects created in RNotebook 2 '2_Soundscape_metrics.Rmd'. 
  • sample_list.rds A list of soundscape objects separated by sampling day created in RNotebook 2 '2_Soundscape_metrics.Rmd'. 
  • inc_mat_island.rds An island-wide OSU-by-sample incidence matrix, created in RNotebook 2 '2_Soundscape_metrics.Rmd'.
  • inc_mat.rds A plot-scale OSU-by-sample incidence matrix, created in RNotebook 2 '2_Soundscape_metrics.Rmd'.
  • unrarefied_gamma_soundrich.rds A dataframe containing information on the unrarefied gamma soundscape richness, as well as associated metadata. Created in RNotebook 2 '2_Soundscape_metrics.Rmd'.
  • rarefaction_cmin.rds A list of rarefaction data for the gamma, alpha and beta soundscape richness for 24 study islands. A coverage-based rarefaction approach to Cmin was used. Data object created in RNotebook 8 '8_Decomposing_SSARs_with_rarefaction.Rmd'. 
  • rarefaction_cmax.rds A list of rarefaction data for the gamma, alpha and beta soundscape richness for 24 study islands. A coverage-based rarefaction approach to Cmax was used. Data object created in RNotebook 8 '8_Decomposing_SSARs_with_rarefaction.Rmd'. 
  • rarefaction_cmin_total.rds A data frame containing aggregatred rarefaction data for 'rarefaction_cmin.rds'. Data object created in RNotebook 8 '8_Decomposing_SSARs_with_rarefaction.Rmd'. 
  • rarefaction_cmax_total.rds A data frame containing aggregatred rarefaction data for 'rarefaction_cmax.rds'. Data object created in RNotebook 8 '8_Decomposing_SSARs_with_rarefaction.Rmd'. 
• QGIS_data
  • Shapefiles: a directory 8 containing useful shapefiles of islands, mainland, reservoir outline, plots locations, and more. Used in RNotebook 3 '3_Map.Rmd' for creating Figure 1, showing a map of the study area. Also used in RNotebook 1 '1_Metadata_preparation.Rmd'  to calculate isolation variable metadata. 
  • Total_buffer_area: a directory containing 40 '.csv' files containing information on the proportion of land in a range of buffer sizes (40 - 2000 m at 50m increments) for each study island. This information was derived in QGIS and is processed in RNotebook 1 '1_Metadata_preparation.Rmd' . 
  • Clipped_buffer_area: a directory containing 40 '.csv' files containing information on the proportion of land in a range of buffer sizes (40 - 2000 m at 50m increments) for each study island. This information was derived in QGIS and is processed in RNotebook 1 '1_Metadata_preparation.Rmd' . 
• Functions: This directory contains some functions that are called in the RNotebook, but do not make part of the core scripts. 

2.3. Detailed file descriptions

A. CVR-index '.csv' files (Data/Index_files)

filename: varied (CVR-index file '.csv' files for each island / sampling plot / sound file combination). Format = G._ISLANDNAME_DATE_TIME

• content: The values of CVR-index values calculated for each sound file. Each '.csv' file contains 128 values (columns), or one per frequency bin (172 Hz width). The unusual file format (value by columns instead of rows) is the result of how indices are calculated and is required for downstream analysis.

• columns:

  • column 1 (Index): No information - ignore.
  • columns 2-129: CVR-index values (proportion of cells in noise-reduced frequency bin with value > 3dB) for each frequency bin (0-172 Hz, 173-344 Hz, etc...)

B. Metadata files (Data/Metadata)

filename: metadata_sites.csv

• content: This metadata file contains the information for all the locations at which acoustic data was collected at the Balbina Hydroelectric Reservoir. For more information on how this data was collected, see Bueno et al. 2020.

• columns:

  • island_name: The name of the island on which acoustic sensors were placed. For more information, see Bueno et al. 2020 (see above).
  • site: The name of the sampling point at which acoustic data was collected. For more information, see Bueno et al. 2020 (see above).
  • latitude: The latitude of the sampling point, in decimal degrees.
  • longitude: The longitude of the sampling point, in decimal degrees.
  • area: The size of the island, in hectares.
  • ndays: The number of full days during which the soundscape was recorded.
  • first_day: The date of the first recording day of the soundscape at each sampling point (YYYY-MM-DD).

• rows: Each row represents a sampling point at which soundscape data was collected, as described in Bueno et al. 2020 (see above).

filename: metadata_clean.csv

• content: This metadata file contains information for the locations used in this study.

• columns:

  • island_name: The name of the island on which acoustic sensors were placed. For more information, see Bueno et al. 2020 (see above).
  • site: The name of the sampling point at which acoustic data was collected. For more information, see Bueno et al. 2020 (see above).
  • latitude: The latitude of the sampling point, in decimal degrees.
  • longitude: The longitude of the sampling point, in decimal degrees.
  • area: The size of the island, in hectares.
  • ndays: The number of full days during which the soundscape was recorded.
  • first_day: The date of the first recording day of the soundscape at each sampling point (YYYY-MM-DD).
  • nplots: The total number of sampling plots at which acoustic data was collected on each island, after removal of corrupted sites.

• rows: Each row represents a sampling point at which soundscape data was collected, as described in Bueno et al. 2020 (see above).

filename: metadata_isolation.csv

• content: This metadata file contains information for the locations used in this study.

• columns:

  • island_name: The name of the island on which acoustic sensors were placed. For more information, see Bueno et al. 2020 (see above).
  • site: The name of the sampling point at which acoustic data was collected. For more information, see Bueno et al. 2020 (see above).
  • latitude: The latitude of the sampling point, in decimal degrees.
  • longitude: The longitude of the sampling point, in decimal degrees.
  • area: The size of the island, in hectares.
  • ndays: The number of full days during which the soundscape was recorded.
  • first_day: The date of the first recording day of the soundscape at each sampling point (YYYY-MM-DD).
  • nplots: The total number of sampling plots at which acoustic data was collected on each island, after removal of corrupted sites.
  • mrb_dist: The buffer distance used (meters) around each island edge to calculate the PW (proportion of water in buffer) isolation variable.
  • proportion_water: The PW (proportion of water in buffer around each island) isolation variable.
  • DNI: The 'distance to nearest island' isolation variable, capturing the edge-to-edge distance (meters) from a focal island to the nearest other island.
  • DMI: The 'distance to nearest mainland' isolation variable, capturing the edge-to-edge distance (meters) from a focal island to the nearest edge of the mainland

• rows: Each row represents a sampling point at which soundscape data was collected, as described in Bueno et al. 2020 (see above).

C. RData files (Data/RData)

filename: balbina_soundscapes.rds

• content: For each plot on each island in the study, this RData file contains a an S4-object (soundscape object) with the information used to calculate the soundscape richness values. For a detailed overview of the structure of a soundscape object, consult this link. Within the soundscape object, calculations are performed on the 'aggregated_df' dataframe, which contains information on how many days OSUs were detected for each time (columns - HH:MM:SS) and frequency bin (rows - upper limit as rownames).

filename: inc_mat.rds

• content: For each plot on each island in the study, this RData file contains an incidence matrix with detection/non-detection data of Operational Sound Units (OSUs - rows) for each sampling day (columns).

filename: inc_mat_island.rds

• content: For each island in the study, this RData file contains an incidence matrix with detection/non-detection data of Operational Sound Units (OSUs - rows) for each sampling day and each plot on that island (columns).

filename: rarefaction_cmax.rds

• content: For each large island in the study (area > 9.4 ha), this RData file contains a list with the outputs of the rarefaction procedure of iNEXT.beta3D to Cmax, including gamma, alpha, and beta values, as well as several dissimilarity measures. For an overview of the outputs of iNEXT.beta3D, consult this link. 

filename: rarefaction_cmin.rds

• content: For each large island in the study (area > 9.4 ha), this RData file contains a list with the outputs of the rarefaction procedure of iNEXT.beta3D to Cmin, including gamma, alpha, and beta values, as well as several dissimilarity measures. For an overview of the outputs of iNEXT.beta3D, consult this link.

filename: rarefaction_cmax_total.rds

• content: This RData file contains a data frame with the combined output of rarefaction to Cmax using iNEXT.beta3D for the gamma (24 islands or rows), alpha (24 islands or rows) and beta (13 islands islands or rows) soundscape richness.

• columns:

  • island_name: The name of the island for which rarefaction results are presented.
  • Order.q: The order of diversity (based on Hill numbers). Here, q=0 means the soundscape richness is presented.
  • SC: The sample coverage to which the rarefaction was performed (here Cmax).
  • Size: The sampling effort (number of sampling days) corresponding to the sample coverage used for rarefaction.
  • value: The soundscape richness value at Cmax (in percentage of total discoverable OSUs).
  • Method: Whether the rarefied values represent the observed values at the sampling effort, or were rarefied/extrapolated.
  • s.e.: The standard error around rarefaction estimates.
  • LCL: The lower 95% confidence interval.
  • UCL: The upper 95% confidence interval.
  • Diversity: The diversity type used by iNEXT.beta3D (here TD refers to 'Taxonomic Diversity' - even though we are working with soundscape values). Consult https://github.com/KaiHsiangHu/iNEXT.beta3D for more information.
  • type: The type of biodiversity that is provided for each row (one of 'gamma', 'alpha' or 'beta).
  • area: The size of the island for which rarefaction results are presented, in hectares.

filename: rarefaction_cmin_total.rds

• content: This RData file contains a data frame with the combined output of rarefaction to Cmin using iNEXT.beta3D for the gamma (24 islands or rows), alpha (24 islands or rows) and beta (13 islands islands or rows) soundscape richness.

• columns:

  • island_name: The name of the island for which rarefaction results are presented.
  • Order.q: The order of diversity (based on Hill numbers). Here, q=0 means the soundscape richness is presented.
  • SC: The sample coverage to which the rarefaction was performed (here Cmin).
  • Size: The sampling effort (number of sampling days) corresponding to the sample coverage used for rarefaction.
  • value: The soundscape richness value at Cmin (in percentage of total discoverable OSUs).
  • Method: Whether the rarefied values represent the observed values at the sampling effort, or were rarefied/extrapolated.
  • s.e.: The standard error around rarefaction estimates.
  • LCL: The lower 95% confidence interval.
  • UCL: The upper 95% confidence interval.
  • Diversity: The diversity type used by iNEXT.beta3D (here TD refers to 'Taxonomic Diversity' - even though we are working with soundscape values). Consult https://github.com/KaiHsiangHu/iNEXT.beta3D for more information.
  • type: The type of biodiversity that is provided for each row (one of 'gamma', 'alpha' or 'beta).
  • area: The size of the island for which rarefaction results are presented, in hectares.

filename: sample_list.rds

• content: This RData file contains a list of 69 elements, one for each sampling plot in the study. Each list element contains five S4-objects (soundscape objects), or one per sampling day. For a detailed overview of the structure of a soundscape object, consult https://thomasluypaert.github.io/soundscapeR_vignette. Within the soundscape object, calculations are performed on the 'aggregated_df' dataframe, which contains information on how many days OSUs were detected for each time (columns - HH:MM:SS) and frequency bin (rows - upper limit as rownames).

filename: unrarefied_gamma_soundrich.rds

• content: This RData file contains a data frame with the unrarefied gamma soundscape richness values for each island in our study (49 island - rows), as well as various types of metadata.

• columns:

  • island_name: The name of the island for which rarefaction results are presented.
  • gamma: the unrarefied gamma soundscape richness value (in percentage of total discoverable OSUs).
  • latitude: the latitude of the study island in decimal degrees.
  • longitude: the longitude of the study island in decimal degrees.
  • area: the size of the island in hectares.
  • proportion_water: The proportion of water in the surrounding buffer (proportion of total buffer area containing water).
  • DNI: The distance to nearest island (in meters).
  • DNM: The distance to nearest mainland (in meters).

\

Sharing/Access information

CVR index data for each sound file was based on the raw sound files available on ARBIMON:

• https://arbimon.org/p/balbina/insights

Code/Software

The RNotebooks described above rely on the following R packages to run:

• rstudioapi : version 0.16.0
• ggplot2  : version 3.5.0
• magrittr : version 2.0.3
• dplyr : version 1.1.4
• ggpubr : version 0.6.0
• sf : version 1.0-15
• data.table : version 1.14.8
• knitr : version 1.45
• kableExtra : version 1.4.0
• purrr : version 1.0.2
• hilldiv : version 1.5.1
• soundscapeR version 0.1.0 (under development on GitHub: github.com/ThomasLuypaert/soundscapeR)
• raster : version 3.6-26
• rgeoboundaries : version 1.3
• elevatr : version 0.99.0
• ggspatial : version 1.1.9
• rworldmap : version 1.3-8
• sars : version 1.3.6
• patchwork : version 1.2.0
• olsrr : version 0.5.3
• iNEXT.beta.3D : version 1.0.1
• ggridges : version 0.5.4

The relationship between RNotebooks is sequential, going from 1 to 10. The relationship between data files, where they are used, and how they relate to eachother is described in section 2 - File descriptions.