Authors

Yuwen Cheng et al.

Description

This dataset contains the raw behavioral data collected from controlled laboratory experiments testing antipredator responses of domestic chicks (Gallus gallus domesticus) to artificial prey bearing eyespot patterns. The experiments examined (1) how partial occlusion of eyespots influences predator behavior (Task 1), and (2) whether prior experience with different eyespot forms alters subsequent responses to a standardized eyespot prey (Task 2).

The data support the analyses presented in the associated manuscript and are intended to enable transparency, reuse, and reanalysis of behavioral responses to antipredator visual signals.

Experimental Design and Methods (Summary)

Domestic chicks were used as a visually guided avian predator model. All individuals were naïve to eyespot stimuli prior to experimental exposure. Chicks were housed under controlled temperature and light conditions and trained to forage along a standardized runway apparatus.

Each experiment consisted of three phases:

1. Adaptation (Days 1–3): acclimation to housing and environment.
2. Training/Habituation (Days 4–6): familiarization with the runway and acquisition of pecking behavior using blank prey (no eyespots).
3. Testing (Day 7): behavioral responses to experimental prey were recorded.

Artificial prey consisted of mealworms mounted on gray triangular paper, bearing black-and-white concentric eyespots. All trials were conducted in a standardized runway with a vertical wooden post and leafy branches to simulate a structured, vegetation-rich microhabitat.

Task 1: Eyespot Occlusion / Microhabitat Visibility

This task tested whether occluding the eyespot signal itself alters antipredator responses. Four treatment conditions were used:

• POE: Partially Occluded Eyespots (eyespot signal partly hidden by a leaf)
• EE: Exposed Eyespots (eyespot fully visible; control for POE)
• WO: Wing Partially Occluded (non-signal wing region partly occluded; eyespot unchanged)
• NO: No Occlusion (entire prey fully visible; control for WO)

Task 2: Experience-Dependent Effects

This task tested whether prior experience with different eyespot forms affects responses to the same concentric-eyespot prey. Four experience conditions were used:

• N: Naïve (no prior exposure to eyespot prey)
• EXP-diff: Experienced with eyespots markedly different from the test stimulus
• EXP-sim: Experienced with eyespots perceptually similar to the test stimulus
• EXP-id: Experienced with identical eyespots used in the test

Task 1 and Task 2 used independent cohorts of chicks and addressed different hypotheses.

Data File Description

File: Supplementary_Materials.xlsx

This Excel file contains the raw behavioral measurements for all tested chicks from both Task 1 and Task 2.

• Each row represents one individual chick tested once.
• Only individuals that completed the testing phase are included.
• No standardized (z-scored) variables are included; all values are raw measurements.

Variable Definitions

Task_name

Experimental condition assigned to the chick.

Possible values include:

• POE – Partially Occluded Eyespots
• EE – Exposed Eyespots
• WO – Wing Partially Occluded (non-signal control)
• NO – No Occlusion (control)
• N – Naïve (Task 2)
• EXP-diff – Experienced with different eyespots
• EXP-sim – Experienced with similar eyespots
• EXP-id – Experienced with identical eyespots

---

prey_type

Label indicating the type of artificial prey presented (e.g., concentric eyespot stimulus).
This variable serves as a stimulus identifier and does not represent a behavioral measure.

---

ID

Unique identifier for each chick, corresponding to numbered leg bands.

---

weight

Body mass of the chick measured prior to testing.
Unit: grams (g)

---

approaching_latency

Time from removal of the visual barrier to the chick’s first entry into the pecking zone.

Unit: milliseconds (ms)

If a chick did not approach within 30 seconds, the latency was right-censored and recorded as 30000 ms.

---

pecking_latency

Time from the chick’s first entry into the pecking zone to the first peck on the prey.

Unit: milliseconds (ms)

If no peck occurred within 30 seconds after entering the pecking zone, the latency was right-censored at 30000 ms.

---

initail_hesitation

Duration of the first pause (hesitation) during the approach toward the prey.

Unit: milliseconds (ms)

---

average_hesitation

Mean duration of all hesitation events observed during the approach phase.

Unit: milliseconds (ms)

---

total_hesitation

Cumulative duration of all hesitation events during the approach.

Unit: milliseconds (ms)

---

tangential_approach

Binary variable indicating the chick’s approach trajectory toward the prey.

• 0 = Direct approach
• 1 = Tangential approach

---

Notes on Data Analysis

In the associated manuscript, continuous behavioral variables were standardized (z-scores) using the R function scale() prior to analysis. Linear mixed-effects models (LMMs) were fitted with experimental condition as a fixed effect and chick ID as a random intercept. However, the present dataset contains only raw, unstandardized values to facilitate flexible reuse.

---

Software and Code

All statistical analyses were conducted in R (version 4.1.3) using the packages dplyr, lme4, and lmerTest. Relevant analysis scripts are provided in the supplementary materials of the associated publication.

File: Rcod_eyespot.R

The R code was used to analyze behavioral responses of chicks using linear mixed-effects models (LMMs). Separate analyses were conducted for Task 1 and Task 2, corresponding to different experimental hypotheses. For each dataset, approach latency, pecking latency, and hesitation time were analyzed independently.

Data import and inspection

data_long <- read.csv("task1data.csv")
str(data_long)
head(data_long)

The raw behavioral data were imported from CSV files. The structure and first rows of the data were inspected to verify variable types and data integrity before analysis.

Data standardization

data_long$approaching_latency <- scale(data_long$approaching_latency)
data_long$pecking_latency <- scale(data_long$pecking_latency)
data_long$average_hesitation <- scale(data_long$average_hesitation)

All continuous behavioral variables were standardized using the R function scale(), which centers each variable to a mean of zero and scales it to unit standard deviation.
This standardization facilitates model convergence and allows effect sizes to be interpreted in standard deviation units.

Linear mixed-effects models

mod1 <- lmer(approaching_latency ~ Task_name + (1 | weight), data = data_long)
summary(mod1)

Linear mixed-effects models were fitted using the lmer() function from the lme4 package.
In each model:

• The behavioral response variable (e.g., approaching latency) was treated as the dependent variable.
• Experimental condition (Task_name) was included as a fixed effect.
• Body mass (weight or bodymass7) was included as a random intercept to account for individual-level variation.

Separate models were fitted for:

• approaching latency
• pecking latency
• average hesitation time

The same modeling structure was applied consistently across all comparisons (POE, EE, POE vs. WO, and Task 2).

---

Task-specific analyses

• Task 1 (POE, EE, and POE vs. WO):
  Models were fitted using weight as the random effect.
• Task 2 (experience treatments):
  Models were fitted using bodymass7 as the random effect, reflecting body mass measured at the testing stage.

Each task was analyzed independently because they involved different experimental cohorts and hypotheses.

Summary

In summary, the R code implements a standard and transparent analysis pipeline:
raw data inspection → variable standardization → linear mixed-effects modeling of continuous behavioral responses → Fisher’s exact tests for categorical approach behavior.