Data for: High contrast markings can negate the benefits of transparent camouflage
Data files
Apr 08, 2026 version files 1.57 GB
-
MashpiFrogs_COLdata.csv
6.70 KB
-
MashpiFrogs_ModelPhotographs.zip
1.57 GB
-
MashpiFrogs_PATdata.csv
42.87 KB
-
MashpiFrogs_PeakFreqdata.csv
1.46 KB
-
MashpiFrogs_SurvivalData.csv
94.70 KB
-
README.md
6.51 KB
-
RmarkdownFiles_vR1.zip
15.15 KB
Abstract
Transparency is, in theory, the ultimate form of concealment allowing for perfect background matching camouflage regardless of the environment. In nature, despite some remarkable examples of highly transparent organisms, physiological constraints mean that transparency is often partial or imperfect. This raises the question of how deviation from true transparency may affect detectability and how camouflage functions. Indeed, it has recently been suggested that partial transparency may function as disruptive camouflage as adjacent transparent and opaque patches differentially blend into the background. Differential blending may therefore offer a route by which obligate opaque structures may be concealed. The glass frogs (Centrolenidae) are a classic example of transparency with ventral skin that allows for a view of the internal organs. However, although the ventral skin is transparent, and the frogs appear translucent, the internal organs are still largely opaque. Here we performed visual modelling and a field predation study with model frogs, to ask how the degree of transparency, and the arrangement of opaque structures, affects detectability and survival. We predicted that greater transparency would improve concealment by facilitating more effective differential blending and that opaque elements which highlighted the recognisable shape of the frog would suffer increased predation risk. We found that greater translucency did improve background matching, but the presence of salient opaque patterns negated this camouflaging effect. Instead, survival was best explained by the distance at which the opaque patterning could be resolved, with thinner edge stripes receiving fewer attacks than thicker central stripes. Our data suggest that although transparency may facilitate effective differential blending, camouflage may be undermined by the presence of obligate opaque structures. These limitations to the efficacy of transparent camouflage may favour the evolution of translucency and explain why, despite having transparent ventral skin, glass frogs retain sparse green pigmentation in their dorsal skin.
Dataset DOI: 10.5061/dryad.xd2547dw6
Description of the data and file structure
Here we include all the raw data from Yeager et al.High contrast markings can negate the benefits of transparent camouflage.
This includes the raw survival data collected in the field, the RAW image files used in the visual modelling, and the processed colour and pattern contrast data derived from the RAW images.
The survival data (MashpiFrogs_SurvivalData.csv) includes the timings of all predation events recorded during the field experiment.
The images files (MashpiFrogs_ModelPhotos.zip) include all images used to quantify model colouration and patterning (n = 20 / treatment). Each model was photographed twice, once in the human visible (VIS) and once in the ultraviolet (UV).
The colour contrast data (MashpiFrogs_COLdata.csv) includes the processed visual modelling data (chromatic and achromatic contrast) for both background matching (model vs leaf) and internal contrast (base vs stripe).
The pattern data (MashpiFrogs_PATdata.csv) includes the processed visual modelling data for the pattern energy of both models and the leaf background at the six spatial scales.
The peak pattern frequency data (MashpiFrogs_PeakFreqdata.csv) includes the processed visual modelling data for the peak spatial frequency from the pattern of each model and leaf.
Contact:
Justin Yeager (Universidad de Las Américas Quito), email: yeagerjd@gmail.com
James B. Barnett (Trinity College Dublin), email: jbarnett@tcd.ie
Location:
Mashpi Reserve, Provincia de Pichincha, Ecuador.
Files and variables
File: MashpiFrogs_SurvivalData.csv
Description: Raw survival data from the field predation study.
Variables
- Round: The three independent replicates of the transect route (A, B, & C).
- Block: Each transect was split into 60 blocks, each of which included 8 models (1 of each treatment).
- StartDate: The date on which models were placed out in the field (0 h).
- EndDate: The date on which the Round ended and model were removed from the field (72 h).
- Treatment: Unique code for each of the four model types (OC = opaque-centre, OE = opaque-edge, TC = transparent-centre, TE = transparent-edge).
- Opacity: Binary code representing the colour of the base model (opaque or transparent).
- Pattern: Binary code representing the pattern of the model (centre or edge).
- Height: Binary code representing the height at which models were placed (high or low).
- Time: The time at which an event (e.g., a predation attempt) was recorded (24, 48, or 72 h)
- Censor: Binary code representing whether an event was categorised as bird predation (1) or as a censored value (0).
- Censor_all: Binary code representing whether an event was categorised as predation (1 - including both confirmed bird attacks and missing models) or as a censored value (0).
- Notes: Description of event characteristics.
File: MashpiFrogs_COLdata.csv
Description: Visual modelling data for colour contrast (chromatic and achromatic).
Variables
- Visual_model: The visual system used for the visual modelling. The Eurasian blue tit (Cyanistes caeruleus) including the LWS, MWS, SWS, UVS, and Double cones.
- Camera: The make and model of the camera used for the visual modelling.
- Photo: Unique alphanumeric code for each individual photograph.
- Treatment: Unique code for each of the four model types (OC = opaque-centre, OE = opaque-edge, TC = transparent-centre, TE = transparent-edge).
- Colour: Binary code representing the colour of the base model (opaque or transparent).
- Pattern: Binary code representing the pattern of the model (centre or edge).
- Replicate: Unique number for each of the 10 replicates of each treatment.
- Mean_COL: Chromatic contrast (dS) between the average colour of the model and the leaf background.
- Mean_LUM: Achromatic contrast (dL) between the average colour of the model and the leaf background.
- Base_COL: Chromatic contrast (dS) between the base colour of the model and the leaf background.
- Base_LUM: Achromatic contrast (dS) between the base colour of the model and the leaf background.
- Stripe_COL: Chromatic contrast (dS) between the stripe colour of the model and the leaf background.
- Stripe_LUM: Achromatic contrast (dL) between the stripe colour of the model and the leaf background.
- Internal_COL: Chromatic contrast (dS) between the base and stripe colours of the model.
- Internal_LUM: Achromatic contrast (dS) between the base and stripe colours of the model.
- Notes: Additional information.
File: MashpiFrogs_PATdata.csv
Description:
Variables
- Visual_model: The visual system used for the visual modelling. The Eurasian blue tit (Cyanistes caeruleus) with pattern energy measured using the Double cone.
- Camera: The make and model of the camera used for the visual modelling.
- Label: Unique alphanumeric code including the region of interest (ROI) and the photograph.
- ROI: The region of interest (ROI) from which data was extracted (frog or leaf).
- Treatment: Unique code for each of the four model types (OC = opaque-centre, OE = opaque-edge, TC = transparent-centre, TE = transparent-edge).
- Colour: Binary code representing the colour of the base model (opaque or transparent).
- Pattern: Binary code representing the pattern of the model (centre or edge).
- pattern_size: The spatial frequency of the Fast Fourier Analysis filter in pixels (22, 44, 88, 176, 352, or 704 px).
- pattern_energy: The energy found at each pattern spatial frequency.
File: MashpiFrogs_PeakFreqdata.csv
Description: The peak spatial frequency (pattern size) for each model and leaf.
Variables
- Label: Unique alphanumeric code including the region of interest (ROI) and the photograph.
- ROI: The region of interest (ROI) from which data was extracted (frog or leaf).
- maxFreq: The spatial frequency of the Fast Fourier Analysis filter in pixels which had the highest energy value (22, 44, 88, 176, 352, or 704 px).
File: MashpiFrogs_ModelPhotographs.zip
Description: Original RAW images (.CR2) of the models used for the visual modelling.
File: RmarkdownFiles_vR1.zip
Description: Rmarkdown files (.rmd) and PDFs (.pdf) for all statistical analyses and plots.
