Dataset DOI: 10.5061/dryad.cz8w9gjkd

Description of the data and file structure

In this study, we tested whether Argentine ants distribute their arolium engagement across legs according to the required normal forces predicted by a quasi-static model. A high-speed camera recorded Argentine ants climbing on a vertical Frustrated Total Internal Reflection (FTIR) sensor. This dataset contains the measured kinematics and arolium engagement status from the experiments.

We tracked ants' six tarsal tips, the anterior end of the head (‘Head’) and the posterior end of the gaster (‘End’) using DeepLabCut. For each frame, the center of body (CoB) position was the middle point between the "Head" and "End". The orientation of the body was defined as the angle between the vector from "End" to "Head" and the gravity direction. All videos were separated into steps (from touch-down to lift-off) for a hindlimb. Over multiple steps, the periodic lowest points of the foot‑body distance were identified as the touch-down moments, while after each touch-down, the first frame when the hind leg moved faster than 16mm/s was treated as the corresponding lift-off. 

Arolium engagement was indicated by the illuminated spot on the tarsal tip generated by the FTIR sensor. We identified arolium engagement by comparing pixel brightness at the foot with respect to the surrounding background noise. Specifically, we used two metrics: (1) a binary representation of whether arolia are engaged/not engaged, which is computed by thresholding the arolium light intensity, and (2) a measure of arolium engagement intensity, which is computed by summing all pixels that exceed the noise threshold.

Files and variables

File: Tripod_coordination_strength.R

Description: R code used to analyze Tripod coordination Strength (TCS). 

File: Engagement_probability.R

Description: R code used to analyze the arolium engagement probability. For uppermost and lowest arolia, we analyzed the engagement status as a function of foot position, climbing direction, and normalized time. For middle arolium, we analyzed the engagement status as a function of stance type, climbing direction, and normalized time. The analysis proceeded as follows: (1) separate binomial regression models were fitted to the engagement state (engaged/ not engaged) for each ant collection; (2) the collection-specific model results were synthesized in univariate meta-analyses; and (3) collection-specific differences between climbing up and down or between wide and narrow stance were calculated and synthesized. 

File: Engagement_intensity_ratio_VS_required_adhesion_force_ratio.R

Description: This R code performs linear regression analysis to study the relation between the engagement intensity ratio and the required adhesion force ratio. The analysis proceeded as follows: (1) a linear regression model (log(engagement_intensity_ratio)~log(required_adhesion_force_ratio)) was fitted to data for each ant collection; (2) the collection-specific model results were synthesized in univariate meta-analyses.

File: Engagement_intensity_VS_time.R

Description: R code used to analyze the temporal variation of engagement intensity. The analysis proceeded as follows: (1) a second-order polynomial regression model was fitted to data for each ant collection (The model is y= a(x-0.5)^2+b(x-0.5)+c, where y is the standardized arolium engagement intensity, x is the normalized time, a is the quadratic coefficient, b is the linear coefficient, and c is the intercept.) and the corresponding vertex (-b/(2a)) was calculated; (2) the collection-specific model results were synthesized in univariate meta-analyses.

File: Acceleration_meta_analysis.R

Description: This R code is used to analyze the temporal variation and the maximum magnitude of acceleration during each step. The analysis proceeded as follows: (1) linear regression models (instantaneous acceleration as a function of normalized time; maximum magnitude of acceleration as a function of climbing direction) were fitted to data for each ant collection; (2) the collection-specific model results were synthesized in univariate meta-analyses.

Shared Variables:

1. Colony (colony): collection name. The letter 'C' stands for 'collection', and the number indicates the month and day (MMDD) of the collection.
2. VName: video name.
3. Nseg: section number. Individual ants could not be reliably identified across the video recordings. Therefore, we treated each continuous walking bout (from entry to exit) as an independent 'section' and numbered them sequentially in order of appearance.
4. Nstride (numStride): stride number.
5. Vframe: video frame number
6. legStance: This indicates in this step which legs are in stance phase. If it is 145, the front left, middle right, and hind left legs are in stance phase. If it is 236, the front right, middle left, and hind right legs are in stance phase.
7. Vdirection: vertical climbing direction. It was defined according to the mean body orientation for each step. If the mean orientation for a step was within the range of 160° to 200°, the climbing direction for this step was defined as "up". If the mean orientation was within the range of 0° to 20° or 340° to 360°, the climbing direction was defined as "down".
8. Phase: normalized time. The normalized time for each frame was calculated according to its position in the whole step frame sequences, with 0 and 1 indicating the beginning and end of a step, respectively.
9. mid_type: stance type. Narrow stance is when the lowest foot is laterally closer to the CoB than the uppermost foot. Wide stance is when the lowest foot is laterally further from the CoB than the uppermost foot. If it is FALSE, the ant is in a narrow stance. If it is TRUE, the ant is in a wide stance.
10. Arolium_position (position): arolium positions were classified as uppermost, middle, or lowest by the arolium's vertical position along the gravity axis.

File: Direction.csv

Description: Body orientation data. 

Variables

• Direction: climbing direction (up/down/not used). If the mean orientation was within the range of 160° to 200°, the climbing direction for this step was defined as "up". If the mean orientation was within the range of 0° to 20° or 340° to 360°, the climbing direction was defined as "down". Other steps were defined as "not used", as they were not used in the further analysis. 
• Mean orientation (unit: degrees (°)): average body orientation for each step. 

File: TCS_data.csv

Description: Data used for Tripod coordination strength (TCS) analysis.

Variables

• legList: hind leg used to separate strides. If it is 1, it is the left hind leg; if it is 2, it is the right hind leg.
• strideStart: the frame number at which the stride starts
• strideEnd: the frame number at which the stride ends
• TCS: TCS value. The TCS values reported here are TCS values for each stride, which were computed as the mean of the TCS values calculated separately for the two steps within the stride. For each step, the TCS was calculated as the ratio between the number of frames in which all three legs of a given tripod were simultaneously in swing phase and the number of frames in which at least one of these legs was in swing phase. For each frame, if a leg's tarsal tip moved faster than 16mm/s, it was treated as in the swing phase, otherwise, it was treated as in the stance phase. We picked this empirical threshold because it is high enough to exceed noise levels yet low enough to represent the transition from stance to swing.

File: GLM_data.csv

Description: Data used to analyze binary engagement status (engaged/not engaged) as a function of climbing direction, leg position, stance type, and normalized time. The corresponding R code is "Engagement_probability.R".

Variables

• engage: This is the binary engagement status, defined by the presence or absence of FTIR illumination. TRUE is engaged (presence of illumination), and FALSE is not engaged (absence of illumination).

File: light_ratio_vs_force_ratio.csv

Description: Data used to analyze the relation between the engagement intensity ratio and the required adhesion force ratio. The corresponding R code is "Engagement_intensity_ratio_VS_required_adhesion_force_ratio.R"

Variables

• light_ratio: The ratio of engagement intensity between two arolia.
• force_ratio: The ratio of required adhesion force between two arolia. The required adhesion force is calculated based on a tripod model. 

Files: up_uppermost_intensity_vs_time.csv, up_middle_intensity_vs_time.csv, up_lowest_intensity_vs_time.csv, down_uppermost_intensity_vs_time.csv, down_middle_intensity_vs_time.csv, down_lowest_intensity_vs_time.csv

Description: Data used to analyze temporal variation of engagement intensity. The corresponding R code is "Engagement_intensity_VS_time.R".

Variables

• strideName: stride name (video name/section number/stride number)
• position: arolium position.
• engagement_intensity: raw engagement intensity (light flux).
• scaleBright: standardized engagement intensity. The engagement intensity was standardized by subtracting its mean and dividing by its standard deviation within each step.

File: acceleration_data.csv

Description: Speed and acceleration data. The corresponding R code is "Acceleration_meta_analysis.R".

Variables

• meanO: mean orientation (unit: degrees (°)). "meanO" requires the addition of 180° to match the mean orientation data reported in "Direction.csv". 
• stepSpeedAlong: instantaneous speed along the body direction (unit: mm/s). The instantaneous speed at each frame was determined by first calculating the displacement between the CoB positions at the previous and next frames. Then, the component of this displacement that aligned with the fore–aft body axis was taken, and divided by twice the frame duration (central difference method).
• stepAcceleAlong: instantaneous acceleration along the body direction (unit: g (9.8 m/s²)). Acceleration was obtained by applying the same central difference method to the instantaneous speed (stepSpeedAlong).

Code/software

All statistical analyses were performed in R (v.4.2.3). Packages used for R code are mentioned in the R code files.