Sex differences in the song circuit and song acoustic complexity in male and female house wrens
Data files
Jul 31, 2023 version files 85.06 MB
-
acoustic.parameters.csv
1.95 MB
-
all.bird.neural.data.csv
20.65 KB
-
all.syllable.acoustic.parameters.w.labels.csv
2.35 MB
-
mds.acoustic.area.points.csv
204.01 KB
-
mds.acoustic.area.points.w.labels.csv
414.19 KB
-
paired.t.test.input.csv
7.55 KB
-
README.md
17.88 KB
-
song.level.data.csv
30.67 KB
-
songs.zip
76.99 MB
-
Transformed.non-colinear.acous.meas.csv
1.41 MB
-
Transformed.non-colinear.acous.meas.w.labels.csv
1.67 MB
Feb 02, 2024 version files 85.06 MB
-
acoustic.parameters.csv
1.95 MB
-
all.bird.neural.data.csv
20.65 KB
-
all.syllable.acoustic.parameters.w.labels.csv
2.35 MB
-
mds.acoustic.area.points.csv
204.01 KB
-
mds.acoustic.area.points.w.labels.csv
414.19 KB
-
paired.t.test.input.csv
7.55 KB
-
README.md
17.45 KB
-
song.level.data.csv
30.67 KB
-
songs.zip
76.99 MB
-
Transformed.non-colinear.acous.meas.csv
1.41 MB
-
Transformed.non-colinear.acous.meas.w.labels.csv
1.67 MB
Abstract
In this study, we compared neural song circuit morphology to singing behavior recorded in the field for 17 male and 18 female house wrens. The acoustic complexity of house wren songs was quantified using a recently published machine learning approach. This data set includes recordings of all house wren songs used in this analysis along with Raven selection tables defining the boundaries of each syllable. This includes 109 female songs. R code used to extract acoustic features and estimate element diversity and our proxy for song acoustic complexity are included. Summaries of acoustic variables for each song and each element are provided as well as files necessary to replicate the analysis. For each bird, we measured volume, cell number, cell density, and neuron soma size for three song circuits, Area X, HVC (used as a proper name), and the robust nucleus of the arcopallium (RA), and one control region, the nucleus rotundus (Rt). This data set includes these neural morphology measurements for each bird as well as R code used to (1) compare males and females for each neural measurement and (2) explore the relationship between acoustic complexity and neural morphology within each sex.
Wild house wrens were recorded in the field singing spontaneously or in response to playback recordings of male or female house wren songs. All songs used in this analysis can be found in the “songs.zip” file. The start and end of each element in the song were defined manually in Raven using both the spectrogram and waveform. These boundaries can be found in the .txt file associated with each sound file (.wav file) in the “songs.zip” folder.
Signal-to-noise ratios (SNR) were used to select songs of suitable quality for the rest of the analysis. Users can use the “snr.and.automatic.frequency.detection.r” script to replicate this calculation for all sounds in the “songs.zip” file. When songs with a suitable SNR were selected, we used this same R script automatically detect the frequency boundaries of each element. These were then viewed in Raven and corrected for any obvious deviations driven by interfering background noise. These final values are included in the .txt file for each sound.
We then used a machine learning approach to quantify acoustic complexity of each song. After transforming and removing any colinear variables, an unsupervised random forest was used to determine which variables best divide the data. This results in a dissimilarity matrix for each syllable which was then transformed into vectors using classical multidimensional scaling. These vectors are “acoustic space” occupied by house wren song elements. A 95% minimum convext polygon was then used to determine how much acoustic space elements within a single song occupy. Songs that occupy more space have a larger range of signal types. This final calculation is referred to as element diversity and is our measure of acoustic complexity.
The “snr.and.automatic.frequency.detection.r” script provides the workflow to replicate this acoustic complexity calculation starting with the songs and .txt files in the “songs.zip” file. Users may also skip earlier steps of this analysis by using the files “acoustic.parameters.csv”, “Transformed.non-colinear.acous.meas.csv” or “mds.acoustic.area.points.csv” as described below.
The “song.level.data.csv” provides summary data on all songs used in the analysis.
17 male and 18 female house wrens were collected, brains were removed, frozen, sectioned, and stained, and neural morphology was measured under brightfield microscopy to quantify neural morphology in three song control regions, Area X, HVC, and RA, and one control region, Rt. Further detailed methods can be found in the associated manuscript. All neural morphology measurements can be found in “all.bird.neural.data.csv”. The “statistics.and.figs.r” script provides the workflow to replicate all statistics and figures in the mansucript. Here we compare males and females for each morphology metric and investigate how song acoustic complexity relates to neural morphology for each sex separately.
Description of the data and file structure
songs.zip: this zipped folder contains songs (.wav files) of each house wren song and selection tables (.txt files) used in the analysis.
-
.wav files: Songs are clipped from much longer recordings with 1 second prior to and 1 second after the boundaries of each song. No additional modification of the sound files occurred. Files are named by song and this notation is used throughout the files in this data set. Bird identity is at the beginning of each name (e.g. 1B.F.1 is song 1 from bird 1B.F).
-
.txt files: Each recording has an associated .txt file that boundaries of each element in the song. This selection table can be opened with the corresponding sound file in Raven to visualize these boundaries on the spectrogram and waveform of the sound. This file is also needed to run the code below that calculates signal-to-noise ratio, automatically detect frequency boundaries, and extract other acoustic parameters.
*Begin Time (s): time element begins in seconds. Time boundaries were set manually using both the waveform and spectrogram with a Hann window with a 50% window overlap and a discrete Fourier transform size of 128 samples.
*End Time (s): time element ends in seconds.
*Low Freq (Hz): the lowest frequency of the element in Hz. These were automatically detected using the “snr.and.automatic.frequency.detection.r” code below and then visualized in Raven and manually corrected for any deviations driven by interfering background noise.
*High Freq (Hz): the highest frequency of the element in Hz. Same methods as above.
song.level.data.csv: a summary document providing information on each song in the data set
- Bird: bird identity
- Sex: bird sex
- Year: year bird was recorded and collected
- Song: individual label for each song. The same labels are used in the songs.zip folder and the script files
- Recording: label for the original recording where each song originated
- Start.time: time within this recording that the song started
- Date: date that each song was recorded
- Stage.recorded: breeding stage during which song was recorded.\
** X eggs: day that egg was laid.\
** nest building: stage prior to egg laying. Birds may be advertising for mates, engaged in courtship, or preparing the nest post-pair-formation.\
** nestlings: bird has dependent nestlings in the nest. post-fledglings: birds has nestlings that recently left the nest (fledged) - Context: context during which song was recorded.\
** spontaneous: bird was recorded singing without the use of artificial playback recordings\
** male.play: bird was recorded while a male house wren recording was playing
** male.post: bird was recorded after it heard male house wren recordings\
** female.play: bird was recorded while a female house wren recording was playing
** female.post: bird was recorded after it heard female house wren recordings - Context.2: context during which song was recorded.\
** m: bird was recorded during or following playback of male house wren recordings\
** f: bird was recorded during or following playback of female house wren recordings\
** s: bird was recorded singing without the use of artificial playback recordings - SNR: signal-to-noise ratio calculated for the song using the “snr.and.automatical.frequency.detection.r” script
- ED.acoustic.space: song element diversity (our proxy for acoustic complexity) calculated using the “acoustic.space.and.ED.calculation.r” script. This is a measure of “acoustic space” occupied by elements in the song
all.syllable.acoustic.parameters.w.labels.csv: acoustic parameters extracted using the ‘specan’ function in the “acoustic.space.and.ED.calculations.r” script with labels added to improve the ease of use.
- sound.files: the sound file which matches the sounds deposited in “song.zip”
- sound.files.syll: a unique label for each element within each sound file. Used as a label throughout the “acoustic.space.and.ED.calculations.r” script workflow
- song: unique identifier for each song. The same labels are used in the “songs.zip” file and throughout the other .csv files
- start: start time of the element in seconds
- end: end time of the element in second
- top.freq: highest frequency of the element in kHz
- bottom.freq: lowest frequency of the element in kHz
- duration: duration of the element in seconds (end-start time)
- meanfreq: mean frequency: average requency weighted by amplitude in kHz
- sd: standard deviation: standard deviation of frequency in kHz
- freq.median: median frequency: frequency in kHz that divides the signal into two intervals of equal energy
- freq.Q25: first quartile frequency: frequency in kHz that divides the signal into 25% and 75% energy
- freq.Q75: third quartile frequency: frequency in kHz that divides the signal in 75% and 25% energy
- freq.IQR: interquartile frequency range: range in kHz between the first quartile frequency and third quartile frequency
- time.median: time in seconds that divides the signal into two intervals of equal energy
- time.Q25: first quartile time: time in seconds that divides the signal into 25% and 75% energy
- time.Q75: third quartile time: time in seconds that divides the signal into 75% and 25% energy
- time.IQR: interquartile time range: range in seconds between the first quartile time and third quartile time
- skew: asymmetry of the spectrum
- kurt: peakedness of the spectrum
- sp.ent: spectral entropy: the frequency spectrum’s energy distribution. Pure tones have low values. Noisy sounds have values close to 1
- time.ent: time entropy: the time envelop’s energy distribution. Pure tones have low values. Noisy sounds have values close to 1.
- entropy: spectrographic entropy: product of time entropy and spectral entopy
- sfm: spectral flatness: N*(prod(y_i)^(1/N)/sum(y_i)) with y = relative amplitude of the i frequency and N = number of frequencies. Similar to spectral entropy. Pure tones have low values. Noisy sounds have values close to 1.
- meandom: average of dominant frequency: average of the frequency of highest amplitude measured across the signal
- mindom: minimum of dominant frequency: lowest value of the frequency of highest amplitude measured across the signal
- maxdom: maximum of dominant frequency: highest value of the frequenc of highest ampliude measured across the signal
- dfrange: range of dominant frequency: range of the frequency of highest amplitude measured across the signal
- modindx: modulation index: cumulative absolute difference between adjacent measurement of dominant frequency/range of the dominant frequency. An unmodulated signal has a value of 1.
- startdom: start dominant frequency: frequency of highest amplitude measured at the start of the signal
- enddom: end dominant frequency: frequence of highest amplitude measured at the end of the signal
- dfslope: dominant frequency slope: (end dominant frequency-start dominant frequency)/duration (kHz/s)
- meanpeakf: mean peak frequency: the frequency from the mean frequency specrum that has the highest energy
acoustic.parameters.csv: data from “all.syllable.acoustic.parameters.w.labels.csv” with labels removed. Users looking the skip the step that requires uploading sound files into R may load this file where prompted in the “acoustic.space.and.ED.calculations” and replicate the rest of the script. Note that subsequent steps will not work if the file with labels is used.
Transformed.non-colinear.acous.meas.w.labels.csv: extracted acoustic parameters that have been centered, scaled, and Box-Cox transformed in preparation for the next step in analysis. Colinear variables have also been removed. Labels are included to improve ease of use.
- same variable labels as “all.syllable.acoustic.parameters.w.labels.csv”. Values have been centered, scaled, and Box-Cox transformed for all variables
Transformed.non-colinear.acous.meas.csv: data from “Transformed.non-colinear.acous.meas.w.labels.csv” with labeled removed. Users looking to skip the transformation step may load this file where prompted in the “acoustic.space.and.ED.calculations” and replicate the rest of the script. Note that subsequent steps will not work if the file with labels is used.
mds.acoustic.area.points.w.labels.csv: Element dissimilarity matrix from the unsupervised random forest transformed through class multidimensional scaling. Each element has a set of two points that places it in “acoustic space”. Elements that are closer together in this space are more acoustically similar. Labels are included to improve ease of use. Users looking to skip the random forest and replicate the element diversity calculations can use this file.
- id: unique identifier for each element. Elements are listed in the same order as they appear in “acoustic.parameters.csv” and “Transformed.non-colinear.acous.meas.csv”
- labels: sound file of the song
- bird: bird identity
- song: song identity. This column is used when calculating average song element diversity.
- sex: bird sex
- X1: first coordinate in acoustic space
- X2: second coordinate in acoustic space
mds.acoustic.area.point.csv: same data as “mds.acoustic.area.points.w.labels.csv” without labels. This file is produced directly by the script “acoustic.space.and.ED.calculations.r”
all.bird.neural.data.csv: a summary document on each bird in the data set used to perform the statistical analyses in the “statistics.and.figs.r” file.
- Bird: bird identity
- Sex: bird sex
- Hemisphere: whether the data that follow come from the left or right hemisphere of the brain
- Year: year bird was recorded and collected
- Stage.captured: breeding stage during which the bird was captured
** prelay: prior to egg laying
** egg: female in the pair laid an egg that day
** incub: female in the pair was sitting on unhatched eggs. female did not lay a new egg that day
** post fledge: offspring from first breeding attempt recently left the nest - Date.capture: date bird was collected
- Yolking.follicle: whether females had yolking follicles when they were collected. This is an indication of active breeding condition.
- ED.acoustic.space.all.songs: average song element diversity for the bird using all songs in the data set. This is the response variable used in the manuscript
- Song.in.acoustic.space.analysis: number of songs from the bird that contributed to the formation of acoustic space in the “acoustic.space.and.ED.calculation.r”. This is the number of songs from each bird in the data set.
- HVC.volume: estimated volume of HVC in mm^3 for that hemisphere
- HVC.cell.count: estimated number of cells in HVC for that hemisphere
- HVC.density: estimated density of cells in HVC (cells/mm^3) for that hemisphere
- HVC.soma.size: average soma size (square micrometers) of 25 neurons for that hemisphere
- HVC.avg.vol: average of HVC volume for both hemispheres. When one hemisphere was missing, just one is used.
- HVC.avg.cell.count: average number of HVC cells for both hemispheres. When one hemisphere was missing, just one is used.
- HVC.avg.dens: average density of HVC cells for both hemispheres. When one hemisphere was missing, just one is used.
- HVC.avg.soma: average HVC soma size for both hemispheres. When one hemisphere was missing, just one is used.
- subsequent columns use the same naming conventions for the other regions (Rt, Area X (X), and RA)
paired.t.test: data from “all.bird.neural.data.csv” re-arranged to facilitate the calculation of paired t-tests in the “statistics.and.figs.r”
- Bird: bird identity
- L.HVC.vol: HVC volume (mm^3) in the left hemisphere
- R.HVC.vol: HVC volume (mm^3) in the right hemisphere
- L.HVC.cell: number of cells in HVC in the left hemisphere
- R.HVC.cell: number of cells in HVC in the right hemisphere
- L.HVC.dens: density of cells (cells/mm^3) in HVC in the left hemisphere
- R.HVC.dens: density of cells (cells/mm^3) in HVC in the right hemisphere
- L.HVC.soma: soma size (square micrometers) of cells in HVC in the left hemisphere
- R.HVC.soma: soma size (square micrometers) of cells in HVC in the right hemisphere
- subsequent columns use the same naming conventions for the other regions (Rt, Area X (X), and RA)
Sharing/Access information
Code in the “acoustic.space.and.ED.calculations.r” file was derived code previously published under a CC BY 4.0 license:
- Keen, Sara (2021): Acoustic diversity dataset. figshare. Dataset. https://doi.org/10.6084/m9.figshare.13661315.v11
Code/Software
Script files are listed in the order they were run. Each script includes internal annotations further describing details of the workflow.
snr.and.automatic.frequency.detection.r: R script to calculate signal-to-noise ratio (SNR) for each song and automatically detect the frequency boundaries of each syllable within the song.
*This script was run in R version 4.1.2.
*required packages: warbleR, Rraven
*required files: sound file (.wav) and associated .txt selection table for each song. Selection tables must share the same name as the .wav files with “selection.txt” added to the end (1B.F.1.wav, 1B.F.1.selections.txt).
acoustic.space.and.ED.calculation.r: R script to calculate acoustic space and estimate element diversity using an unsupervised random forest approach. The resulting measurement is our measure of song acoustic complexity in this manuscript.
*This script was run in R version 4.1.2.
*required packages: parallel, vegan, bioacoustics, warbleR, Rraven, cluster, Rtsne, randomForest, MASS, fossil, pbapply, adehabitatHR, caret, DescTools
*required files: users many run the script from the beginning by using the sound file (.wav) and associated .txt selection table for each song. We have also provided files created during the script workflow for users that would like to replicate portions of the code (acoustic.parameters.csv, Transformed.non-colinear.acous.meas.csv, mds.acoustic.area.points)
statistics.and.figs.r: R script to recreate the statistical analysis and figures in this manuscript.
*This script was run in R version 4.1.2.
*required packages: car, effsize, Hmisc
*required files: “all.bird.neural.data.csv”, “paired.t.test.input.csv”
Wild house wrens were recorded in the field singing spontaneously or in response to playback recordings of male or female house wren songs. Songs were clipped from much longer song recordings with 1 second before the start and 1 second after the end of the song. No further processing occurred. All songs used in this analysis can be found in the "songs.zip" file. The start and end of each element in the song were defined manually in Raven using both the spectrogram and waveform. These boundaries can be found in the .txt file associated with each sound file (.wav file) in the "songs.zip" folder.
Signal-to-noise ratios (SNR) were used to select songs of suitable quality for the rest of the analysis. Users can use the "snr.and.automatic.frequency.detection.r" script to replicate this calculation for all sounds in the "songs.zip" file. When songs with a suitable SNR were selected, we used this same R script to automatically detect the frequency boundaries of each element. These were then viewed in Raven and corrected for any obvious deviations driven by interfering background noise. These final values are included in the .txt file for each sound.
We then used a machine-learning approach to quantify the acoustic complexity of each song. After transforming and removing any colinear variables, an unsupervised random forest was used to determine which variables best divide the data. This results in a dissimilarity matrix for each syllable which was then transformed into vectors using classical multidimensional scaling. These vectors are "acoustic space" occupied by house wren song elements. A 95% minimum convex polygon was then used to determine how much acoustic space elements within a single song occupy. Songs that occupy more space have a larger range of signal types. This final calculation is referred to as element diversity and is our measure of acoustic complexity. The "snr.and.automatic.frequency.detection.r" script provides the workflow to replicate this acoustic complexity calculation starting with the songs and .txt files in the "songs.zip" file. Users may also skip earlier steps of this analysis by using the files "acoustic.parameters.csv", "Transformed.non-colinear.acous.meas.csv" or "mds.acoustic.area.points.csv" as described in the "README.md" document.
17 male and 18 female house wrens were collected, brains were removed, frozen, sectioned, and stained, and neural morphology was measured under brightfield microscopy to quantify neural morphology in three song control regions, Area X, HVC, and RA, and one control region, Rt. Further detailed methods can be found in the associated manuscript. All neural morphology measurements can be found in "all.bird.neural.data.csv". The "statistics.and.figs.r" script provides the workflow to replicate all statistics and figures in the manuscript. Here we compare males and females for each morphology metric and investigate how song acoustic complexity relates to neural morphology for each sex separately.
R can be used to run the code files ("acoustic.space.and.ED.calculations", "snr.and.automatic.frequency.detection", "statistics.and.figs"). These files can also be opened in Notepad. Txt files associated with each sound file can be opened in Raven Pro or Raven Lite (free version) to visualize the boundaries of each element type. These files can also be opened in any program that can read .txt files.