Data from: Vocal characteristics of distress and reproductive vocalizations in North American wapiti
Data files
Jan 28, 2025 version files 102.08 KB
-
README.md
8.25 KB
-
Wapiti_Averaged_Data.xlsx
17.65 KB
-
Wapiti_Raw_Data.xlsx
76.18 KB
Abstract
Variation in the vocal behavior of nonhuman vertebrates includes graded transitions and more dramatic changes. Wapiti males produce a reproductive bugle that has a fundamental frequency that surpasses 2,000 Hz with evidence of biphonation and other nonlinear phenomena. Here, we analyse the acoustic structure of captive wapiti vocalizations to compare the male bugle with three categories of distress vocalizations: neonate distress (capture) calls, calf isolation calls, and adult female isolation calls. These four high-arousal call categories serve a common general function in recruiting conspecifics but occur in different behavioural contexts (capture, isolation, reproduction). Our goal was to distinguish characteristics that vary in graded steps that may correspond to an animal’s age or size from characteristics that are unique to the bugle. Characteristics of the high and loud fundamental (G0) varied in an age/size-graded manner with a decrease in minimum G0, an increase in the maximum and range of G0, with no evidence of sex differences. The nonlinear phenomena of deterministic chaos, biphonation, and frequency jumps were present in all four call categories and became more common from the distress vocalizations of neonates to calves to adult females to the male bugle. Two temporal characteristics sharply distinguished the bugle from the three categories of distress vocalizations: these included a prolonged call duration and a maximum G0 that occurred much later in the call for the bugle than for distress vocalizations. Our results suggest that distress vocalizations of different age groups and the reproductive bugle of wapiti share a high G0, with age/size-graded changes in G0 and nonlinear phenomena, but differ sharply in temporal characteristics.
Corresponding author: Susan Lingle: Department of Biology, University of Winnipeg, Winnipeg, MB. Email: lingle.uw@gmail.com
This README file describes data for two files.
The first file, “Wapiti_Raw Data.xlsx”, includes raw data used to calculate means, standard deviations, and statistical results for a mixed model ANOVA.
The second file, “Averaged Data.xlsx”, includes data that were averaged for each individual before calculating the proportion of calls with each form of non-linear phenomena and running Kruskal-Wallis tests on non-linear phenomena. This data file was also used to conduct a Principal Component Analysis using the non-linear phenomena and a Descriptive Discriminant Analysis (DDA) using the averaged acoustic data for each individual.
FILENAME 1: Wapiti_Raw Data.xlsx
This Excel file has raw acoustic data for the wapiti, plus a few columns with transformed values. Data in this file were used to calculate the mean, SD, and ANOVA results. The file includes columns with data for a few variables that were not included in the final paper.
Column headings: description
Call category: This column represents the four call categories: neonate distress calls (1); calf isolation calls (2); adult female isolation calls (3); and adult male bugles (4).
indiv: Each distinct entry identified a distinct individual. Each individual belongs to only one call category.
sex: Male (1) or Female (2).
Age (days): Represented in days since birth. These were calculated from the animal’s birthdate in the case of neonates and 3-5 month old calves. Birthdates were available to the year for all adults, and converted to days to be consistent with the younger individuals in the data set.
Call ID: This code is specific to a particular vocalization, with the first number representing a unique individual.
The following acoustic values were calculated as described in the paper.
sig-noise (Signal-to-noise)
Call dur (s): Call duration (s)
Log10 call dur: Log10 of call duration: Transformation was “Log10 (Call duration+1)”
TimeMaxG0_calldur: The time of the maximum G0 expressed as a proportion of the call duration.
SqRtTimeG0MaxofCallDur: The transformation for “TimeMaxG0” was “SqRt (TimeMaxG0_Call duration)”
Mean G0 (Hz): the average G0 during the call.
Max G0 (Hz): the maximum G0 during the call.
Min G0 (Hz): the minimum G0 during the call.
Range G0 linear (Hz): The range of G0 expressed as a linear measurement: the maximum G0 minus the minimum G0.
Log10RangeG0linear: Transformation for the Range G0 linear (Hz) was “Log10 (rangeG0linear +1)”
Range G0 Mult: The range expressed as a multiple: the maximum G0 divided by the minimum G0.
Inverse root rangeG0mult: A transformation of the Range G0 Multi. This was calculated as “(1/(sq root(range G0mult))”
HNR whole call: Harmonic Noise ratio based on measurement of the full call duration
HNR slice: Harmonic Noise ratio based on measurement of a spectral slice. This column of data was used for analyses presented in the paper.
Peak frequency slice: the dominant frequency when measuring a spectrum produced from the spectral slice. This column of data was used for analyses presented in the paper.
Peak freq whole call: the dominant frequency based on measuring a spectrum produced from the entire call.
Subharm: Presence (1) or absence (0) of subharmonics during the call
Bipho: Presence (1) or absence (0) of biphonation during the call
DetChaos: Presence (1) or absence (0) of deterministic chaos during the call
Freq Jump: Presence (1) or absence (0) of either ascending or descending frequency jumps during the call.
nlp: The presence of any of the above forms of non-linear phenomena during the call.
EQ25, EQ50, EQ75: The frequency (Hz) of each of three energy quartiles during each call. Data for EQ 25 were presented in the paper.
Mean F0 (Hz): the average value for the F0 in calls with biphonation. When no value is listed, the call usually did not have any visible evidence of biphonation, as indicated by a “0” in the “Bipho” column. However, there were six calls for which there was visible evidence of biphonation via sidebands (entered as a “1” in the “Bipho” column), but we could not reliably measure a value for F0. This included one neonate, three adult female, and two adult male calls. The reasons ranged from the F0 being too fleeting, too low in amplitude, or too inconsistent for us to measure a value without ambiguity based on review of the spectrogram, spectrum, and oscillogram.
Visible_F0_peak: Was the F0 visible in the spectrum? 1, yes; 0, no. If the cell was empty, we detected no biphonation in the call.
FILENAME 2: Wapiti_Averaged Data.xlsx
This file has data that have been averaged for each individual, so that it is suitable for calculating the proportion of calls having each form of non-linear phenomena, and is also suitable for the Principal Component Analysis and the Descriptive Discriminant Analysis. Most column headings include “Ave” to indicate they represent averaged values for one individual.
Column headings: description
Call category: This column represents the four call categories: neonate distress calls (1); calf isolation calls (2); adult female isolation calls (3); and adult male bugles (4).
Individual: Each distinct entry and row identifies a distinct individual. Each individual belongs to only one call category.
Sex: Male (1) or Female (2).
Age (day): Represented in days since birth. These were calculated from the animal’s birthdate in the case of neonates and 3-5 month old calves. Birthdates were available to the year for all adults, and converted to days to be consistent with the younger individuals in the data set.
The following acoustic values were calculated as described in the paper. This file includes an average value for each individual for each of the following traits.
A. Continuous variables:
Ave Call dur (s): The average call duration (s).
Log10 AveCallDur: Log10 of the average call duration: Transformation was “Log10 (Call duration+1)”
Ave TimeMaxG0_calldur: The average time of the maximum G0 expressed as a proportion of the call duration.
SqRtAveTimeG0MaxofCallDur: The transformation for AveTimeMaxG0_calldur was “SqRt (TimeMaxG0_Call duration)”
Ave Max G0 (Hz): The average maximum G0 during the call.
Ave Min G0 (Hz): The average minimum G0 during the call.
Ave Range G0 Mult: The average range of G0 expressed as a multiple: the maximum G0 divided by the minimum G0”.
Inverse root AveRangeG0Mult: A transformation of the Range G0 Multi. This was calculated as “(1/(sq root(range G0mult)).”
AVE HNR slice: The average Harmonic Noise ratio based on measurement from a spectral slice.
Ave Peak frequency slice: The average dominant frequency when measuring a spectrum produced from the spectral slice.
Ave EQ25: The average frequency (Hz) of the first energy quartile, 25th percentile.
B. Non-linear phenomena (NLP): each of the following NLP is expressed as an average proportion of calls with a certain trait for a specific individual.
PROP SH: The average proportion of calls with subharmonics.
PROP BP: The average proportion of calls with biphonation.
PROP CH: The average proportion of calls with deterministic chaos.
PROP FreqJump: The average proportion of calls with either ascending or descending frequency jumps.
PROP NLP: The average proportion of calls with any of the above forms of non-linear phenomena.
C. Other columns:
PC1 NLP: Values for the first Principal Component based on the four forms of non-linear phenomena: subharmonics, biphonation, deterministic chaos, and frequency jumps.
PC2 NLP: Values for the second Principal Component based on the four forms of non-linear phenomena: subharmonics, biphonation, deterministic chaos, and frequency jumps.