Data from: Building variation in visual displays through discrete modifications of motion
Data files
Oct 08, 2025 version files 35.29 MB
-
Fig1_Phylogeny.zip
1.98 MB
-
Fig3_Rates.zip
13.88 KB
-
Fig5_Morphospace.zip
19.06 MB
-
FigS4_Habitat.zip
14.23 MB
-
README.md
6.80 KB
Abstract
Interactions between conspecifics are often composed of one or more discrete behavioural displays. Here we evaluate a behaviour used in aggressive interactions between conspecific males of 10 species of leaf warblers (Phylloscopus). Using high-speed videography and methods derived from geometric morphometrics, we find that the form of a primary visual display differs significantly among species, but with large intraspecific variation and much overlap in shape space. Additional interspecific differences include a species which does not move its wings at all, two quantitatively different displays in the behavioural repertoire, and the loss or gain of a pale patch on the wing. We conclude that display evolution proceeds largely by adding or subtracting discrete components from an established repertoire, accompanied by slight modifications of the core display. In these ways, more complex displays evolve on the background of the ancestral signal, thereby enabling modifications to appear without loss of ancestral efficacy.
Dataset DOI: 10.5061/dryad.1rn8pk177
Description of the data and file structure
Files and variables
For each figure and corresponding analyses there is a folder (zip file) that includes the code and related files. All compressed archives will expand to the folder structure containing: code, data, figures and output.
- Code is where the R or R markdown scripts are located. In Fig1_Phylogeny and Fig3_Rates the R Markdown scripts are located in the main folder.
- Data contains csv files or txt files necessary to run the analyses.
- Figures is where the final figures are saved to.
- Outputs contain several subfolders labeled by file extension. For example, graphs or other figures generated by the code can be saved in the tifs or jpegs folders, while the results of an analysis would be saved in the csv folder.
Note that several of these analyses use a phylogeny. The Phylloscopidae phylogeny (TPPhylltree in the code) is from Alström et al. 2018 and is not included. Please obtain it from the source and place it in data folder associated with the script that is being run. (https://doi.org/10.1016/j.ympev.2018.03.031).
File: Fig1_Phylogeny.zip
Description:
2025_Fig1_Phylogeny.Rmd: R markdown file with the code for generating Figure 1
Warbler_Data_2.csv: Contains columns for species common name, species scientific name (Phylloscopus abbreviated to P), name in phylogeny (how a species name is formatted in the tree file), wing bar size real (in millimeters, from Price and Pavelka 1996) wing bar size (wing bar size divided by 10 for figure formatting), wing bar presence (categorical variable indicating whether birds possess a visible wing bar, P for present, A for absent), primary breeding habitat, nudge factor (figure parameter adjusting placement of trait data on the x-axis of the phylogeny for better visualization), and the species’ wing display repertoire, represented by three columns (Double, Single, Shiver), each coded as present (1) or absent (0).
File: Fig3_Rates.zip
Description:
2025_Fig3_Rates.Rmd, R markdown file with the code for generating Figure 3 and the analyses related to wing flick rates.
Under the data folder: Display_Rates_2024.csv. Contains columns for video identifier (species_individual_date), species, individual, date of video, category of wing flick type, context (whether the bird in video is foraging or responding to territorial playback experiment), raw video name, clip duration (in seconds, length of video in which bird is performing display), WF (number of wing flicks performed), WFS (wing flicks per second, WF divided by clip duration), and primary breeding habitat.
File: Fig5_Morphospace.zip
Description:
This contains two subfolders, Double_WF and All_WF, which correspond to panel A and panel B, respectively, for figure 4. The structure of each folder is the same, which includes an R markdown file with the code, a folder with the shape files (txt files used to generate the wing flick shapes) and a csv file (Mid_Shape_Factor) with the metadata for the shape files. Both metadata csv files (Mid_Shape_Factor_T1_V2.csv and
Mid_Shape_Factor_TAll.csv) have columns with species, individual (re-numbered for each species), raw video name, wing flick coded in numbers (wf, coded as 1 for double, 2 for shiver, 3 for single), wing flick (behavior written out as “single”, “double”, etc) and beak orientation (L if the bird if facing left in the frame of view, R if the bird is facing right, used for orienting the shape files).
File: FigS4_Habitat.zip
Description: contains a folder called "Combined" with the raw data from light logger measurements and 2024_FigS5_Habitat.Rmd, a R markdown file with the code for processing the data and generating supplementary figure 4. The folder Combined_CSV contains data readouts from each light logger saved as a csv file. These files follow the naming convention: TreeType_SampleSite#_LoggerID_DateTimeofCollection. For example, Birch-1-21081620 2022-06-04 14_16_32 IST (Data IST).csv refers to a logger placed in a birch forest, sample site 1, logger ID 21081620, collected on June 4, 2022 at 14:16 India Standard Time. Note that with the exception of Oak and Rhodo (rhododendron) samples, other forest types will have duplicates of the sample site number ID. For example, the dataset contains Birch-1-21081603 2022-07-08 00_16_32 CDT (Data IST).csv and Birch-1-21081620 2022-06-04 14_16_32 IST (Data IST).csv. This is because sample site numbers were reassigned when we moved between our two field sites, Manali Wildlife Sanctuary in Himachal Pradesh, India (32.25°N, 77.17°E), and Nainghar village, Himachal Pradesh, India (32.73°N, 76.86°E). These sample sites can be differentiated by the date they were collected. Loggers from Manali Wildlife Sanctuary were collected by June 4, 2022 and loggers from Naingahar were collected by July 8, 2022.
File: Supplementary_Videos.zip
Description: The Supplementary Videos are deposited via Zenodo and contain 8 samples of videos that depict the behaviors analyzed in the manuscript. We have included examples of both standard frame rate (60 fps) and high frame rate (480 fps) videos where applicable. We chose a representative species for each of the wing flick types (double, shiver, and single) but different individuals are shown in the 60 fps and 480 fps videos.
-
Supp. Video 1: Chick begging display, P. trochiloides, standard frame rate (60 fps).
-
Supp. Video 2: Double wing flick display, P. chloronotus, high frame rate (480 fps).
-
Supp. Video 3: Double wing flick display, P. chloronotus, standard frame rate (60 fps).
-
Supp. Video 4: Shiver wing flick display, P. pulcher, high frame rate (480 fps).
-
Supp. Video 5: Shiver wing flick display, P. pulcher, standard frame rate (60 fps).
-
Supp. Video 6: Single wing flick display, P. occipitalis, high frame rate (480 fps).
-
Supp. Video 7: Single wing flick display, P. occipitalis, standard frame rate (60 fps).
-
Supp. Video 8: P. affinis singing and approaching speaker, standard frame rate (60 fps).
Code/software
All analyses and figures were generated using R version 4.2.2
The required packages are all included in each R markdown file in the corresponding figure folders.
Access information
Data was derived from the following sources:
The phylogeny used in analyses was derived from the following source
We induced aggressive displays using song playback, filmed the displays using high-speed video, and adapted methods from geometric morphometrics to ask the extent to which the form of the display is conserved or varies across species. Using these videos, we also collected information on wing flick rate. The data was processed using QuickTime player to measure rates, and all further analyses conducted in R. We also measured habitat illuminance in the primary breeding habitats across the elevational gradient using Onset light and temperature loggers. Data from the loggers were downloaded to the HOBOconnect app via Bluetooth at the time of collection and analysed using scripts written for R.