Data from: Global selection on insect antipredator coloration
Data files
May 20, 2026 version files 5.42 MB
-
BirdSurveyLocations_2026.xls
43.01 KB
-
dataset_survival_tot_2026_excel_dryad.xlsx
3.03 MB
-
dataset_unpalatable_experiment_2026.xlsx
85.98 KB
-
Location_level_data_GACN_2025.csv
5.64 KB
-
mealworm_analysis.csv
24.83 KB
-
README.md
16.39 KB
-
SuppDataLocationsMaxNumSpecGen_2025B.xls
2.21 MB
Abstract
Natural selection has repeatedly led to the evolution of two alternative antipredator color strategies – camouflage to avoid detection and aposematism to advertise unprofitability – but we lack understanding of how ecological context favors one strategy over the other. We conducted a globally replicated predation experiment at 21 sites on six continents to test how predator community, prey community, and visual environment influenced predation risk of 15,018 paper ‘moth’ artificial prey with cryptic or warning coloration. Results indicated that aposematic strategies fare better in low predation intensity environments, while camouflage strategies are advantaged when other camouflaged prey species are rare and when light levels are low. This study demonstrates how multiple mechanisms shape antipredator strategies, helping explain the evolution and global distribution of camouflaged and aposematic animals.
https://doi.org/10.5061/dryad.8931zcs0j
Predation dataset for global experiment across 21 locations using three types of palatable targets (mealworms). We also present an alternative version of the experiment using unpalatable prey. All analyses are attached in the code.
Description of the data and file structure
Main dataset:
dataset_survival_tot_2026_excel_dryad.xlsx
This dataset contains all the raw predation data on whether targets were attacked or censored and the survival times (15,018 targets). All locations are included and description of each of the columns has been added in a metadata sheet and below. The datafile contains empty cells because not all targets were photographed, only the photographed targets have values in photo-dependent measurements and only targets with multiple photos through the day have data on variation in photo measurements between photos.
| Variable | Description |
|---|---|
| id | Unique identification for each target with location and block information |
| Location | Location were the experiment was performed |
| TargetID | Individual target ID that was created by the local team |
| Treatment | Treatment colour of the target: A= Aposematic, B=Atypical warning colour, C=Camouflage |
| BlockNumber | Day that the event data was collected |
| SurvTime | Time of the day when the target was either taken or considered to be censored |
| Status | 1= taken, 0= censored |
| Censored_Streamlined | Few categories for censoring: Survived or whether taken by a non-avian predator |
| Censored_Original | Original information that came with each target |
| date | Date when the block was run |
| PhotoID_start | ID of photo at beginning |
| illum_start | Illumination of start photo |
| sal_NS_start | Salience of start photo |
| PhotoID_noon | ID of photo at noon |
| illum_noon | Illumination at noon photo |
| sal_NS_noon | Salience at noon photo |
| PhotoID_sunset | ID of photo at sunset |
| illum_sunset | Illumination at sunset photo |
| sal_NS_sunset | Salience at sunset photo |
| PhotoID_sunrise | ID of photo at sunrise |
| illum_sunrise | Illumination at sunrise photo |
| sal_NS_sunrise | Salience at sunrise photo |
| predation_rate | Predation rate calculated from last mealwork block |
| temp_seasonality | Temperature seasonality calculated from bioclim variable |
| blue_species_count | Number of species categorised as blue |
| blue_species_percentage | Percentage of species categorised as blue |
| cryptic_species_count | Number of species categorised as cryptic |
| cryptic_species_percentage | Percentage of species categorised as cryptic |
| warning_species_count | Number of species categorised as warning |
| warning_species_percentage | Percentage of species categorised as warning |
| ratioCrypticWarning | Number of species categorised as cryptic divided by species classified as warning |
| ratioWarningCryptic | Number of species categorised as warning divided by species classified as cryptic |
| log_mean_illum | Log of the mean of the illumination at start photos for each location |
| sd_log_illum | standard deviation across photos at noon for each location |
| mean_sal | mean of the salience at noon photos for each location |
| sd_sal | standard deviation of salience across photos at noon for each location |
| n | Number of photos used to calculate the mean values of illumination and salience per location |
| log_mean_illum_noon | Log of the mean of the illumination at noon photos for each location. THIS IS THE VARIABLE THAT WAS USED |
| local_cv_illum | standard deviation across photos at noon for each location |
| TransectLenghtKm | Transect length for each location |
| InsectivoreAbundance | Abundance but only for species classified as insectivores |
| InsectivoresPerKm | Insectivore species standarised for transect length. We use this variable as density of insectivorous predators. |
| InsectivoreSpeciesNumber | Number of insectivore species |
| InsectivoreShannonDiversity | Shannon diversity index for insectivore species only. We use this variable as predator diversity. |
| Latitude | Latitude of location |
| Sp60ProIndiv | Specialisation index where species are considered specialists with 60% of insect prey in diet but considering the individuals |
| cv_illum_time | variation in illumination across photos for each target |
| log_cv_sal_time | variation in salience across photos for each target (log transformed) |
| match_illum | value of illumination for the last photo before the target was taken (i.e. if target taken at noon, then it would be the start photo) |
| match_sal | value of salience for the last photo before the target was taken (i.e. if target taken at noon, then it would be the start photo) |
| minutes | Minute of the day when the status check was done and when target was considered taken or censored |
Second dataset:
dataset_unpalatable_experiment_2026.xlsx
This dataset contains the raw predation data for the alternative experiment in the Czech Republic that used unpalatable mealworms. The column description is the same as in the main dataset (variables described above). The data file contains empty cells because not all targets were photographed; only the photographed targets have values in photo-dependent measurements, and only targets with multiple photos through the day have data on variation in photo measurements between photos.
code_times
Concatenated value of location and time of check
location
location and type of experiment (CZ_unpalatable is the unpalatable version and Czech_Republic, with palatable prey, same value as in large dataset).
Palatability
U - Unpalatable P- Palatable
TargetID
ID of the target (treatment and number)
Treatment
Treatment colour of the target: A= Aposematic, B=Atypical warning colour, C=Camouflage
BlockNumber
Day that the event data was collected
Survtime
Time of the day when the target was either taken or considered to be censored
Status
1= Taken 0=Censored
StatusB
Censoring those targets that had been taken during night
minutes
Minute of the day when the status check was done and when the target was considered taken or censored
Third dataset:
Location_level_data_GACN_2025.csv
This dataset contains the location-level information for each of the 21 locations. Latitude and longitude used for the plots is also included (lat_b, Lon_b). Column description is the same as in the metadata sheet in the main dataset (also presented above), and values are also the same but summarised per locality.
Fourth dataset:
BirdSurveyLocations_2026.xls
This dataset compiles all the information from the bird surveys in each location and the indices and variables calculated from the bird survey data that were used in posterior analyses. The column description is in a second sheet in the same file, and below:
Location: Location names.
TransectLenghtKm: Lengths of transects as reported by location teams.
TotalBirdAbundance: Total abundances of all birds recorded at particular locations calculated following the BTO Breeding Bird Survey protocol (Newson et al. 2008) as maximum total counts of each species from the two bird surveys carried out.
TotalBirdSpeciesNumber: Numbers of all bird species recorded at particular locations.
PredatorAbundance: Total abundances of birds recorded at particular locations that are classicied as at least partly insectivorous predators. The classification is based on the Elton Traits database (Wilman et al. 2014). This metric filters out the species that are unlikely to be predators of experimental targets (such as specialised granivores or aerial hawkers).
PredatorSpeciesNumber: Numbers of at least partly insectivorous bird species recorded at particular locations.
PredatorsPerKm: Abundances of at least partly insectivorous birds recorded at particular locations standardised per 1 km of the transect.
PredatorShannonDiversity: Shannon Diversity Index based on abundances of at least partly insectivorous species.
Insectivore60ProIndiv: Proportions of bird individuals recorded at particular locations that belong to specialist insectivorous species based on the criterion of having at least 60% of the diet consisting of invertebrates according to Elton Traits database (Wilman et al. 2014).
Team Experience: Team experience with bird identification in the field based on self-assessment (four-grade scale: 1 = highly expert, 4 = familiar with common species).
SurveyObservers: Number of observers that carried out the bird survey.
Fifth dataset:
SuppDataLocationsMaxNumSpecGen_2025B.xls
This dataset contains the raw information from the bird survey experiments with the names of all species detected in each of the locations. The column description is in a second sheet in the same file and below:
Location: Location names.
BinomialName: EltonTraits Binomial names of species as they appear in the Elton Traits database (Wilman et al. 2014). The species names are unified across all locations, with synonyms included in brackets.
MaxNumber: Total species abundances at each location calculated following the BTO Breeding Bird Survey protocol (Newson et al. 2008) as maximum total counts from the two bird surveys carried out.
AvianPredator: Classification of avian species as at least partly insectivorous predators based on the Elton Traits database (Wilman et al. 2014) to filter out the species that are unlikely to be responsible for the predation of experimental targets (such as specialised granivores or aerial hawkers).
Insectivore60: Classification of species as insectivorous specialists based on the criterion of having at least 60% of the diet consisting of invertebrates according to Elton Traits database (Wilman et al. 2014).
Mealworm analysis:
mealworm_analysis.csv
Supplementary dataset with descriptive information on the size of the mealworms used per block and the weather description.
location: Location where data was collected
block: Number of day of data collection
date: Date of data collection
sunrise: Sunrise time
sunset: Sunset time
mealworm_length: average length of mealworms for that block in mm
start_time: time of first deployment
start_weather: weather at first deployment
start_orig_weather: weather in detail (original)
noon_time: time of check at noon
noon_weather: weather at second check
noon_orig_weather: weather at second check (original)
sunset_time: time of check at sunset
sunset_weather: weather at third check
sunset_orig_weather: weather at third check (original)
sunrise_time: time of last check at sunrise
sunrise_weather: weather at last check
sunrise_orig_weather: weather at last check (original)
Code/Software
R markdown files with the description of the analyses and the instructions to build the plots are presented.
Main analyses were generated using the GACN_final_code_streamlined.rmd
Additional files:
Videos of predators attacking and consuming mealworms from the targets presented.
We used a standard and well-established field predation paradigm where triangular paper ‘moths’ printed with different color patterns and attached to a mealworm prey are exposed to free-living predators in natural conditions. We deployed three treatments: 1.) typical warning color pattern consisting of orange and black stripes surrounded by a black border, 2.) atypical warning color pattern that was identically patterned and equally conspicuous (see Stimuli) but with the orange replaced by a turquoise color atypical of warning coloration, and 3.) a cryptic pattern consisting of a fine orange and black checkerboard with the same average color and lightness as the typical warning treatment, but having a brown appearance. To ensure all predators globally would be required to make similar generalizations about the warning treatments’ potential defenses, the warning color pattern was designed to have some general features typical of many aposematic lepidopteran species found globally, but not closely mimic any specific species. Based on effect sizes of previous experiments using this paradigm, anticipated predator learning rates, and how many targets could be placed and checked by experimenters in the time available, at each location, 720 targets (240 per treatment, 90 per day) were tested.