READ ME File For 'Kinematic performance declines as group size increases during escape responses in a schooling coral reef fish' Dataset DOI: https://doi.org/10.5258/SOTON/D2777 Date that the file was created: January 2023 ------------------- GENERAL INFORMATION ------------------- ReadMe Author: Lauren E Nadler, University of Southampton [ORCID ID 0000-0001-8225-8344] Date of data collection: 11/2014 - 08/2023 Information about geographic location of data collection: Field work conducted at the Lizard Island Research Station at Lizard Island, Australia; video data analyses conducted at Nova Southeastern University in Dania, Florida, USA and University of Southampton in Southampton, UK. -------------------------- SHARING/ACCESS INFORMATION -------------------------- License: CC-BY This dataset supports the publication (and recommended citation for the data): AUTHORS: Monica D Bacchus, Paolo Domenici, Shaun S Killen, Mark I McCormick, Lauren E Nadler TITLE: Kinematic performance declines as group size increases during escape responses in a schooling coral reef fish JOURNAL: Frontiers in Fish Science PAPER DOI: 10.3389/frish.2023.1294259 -------------------- DATA & FILE OVERVIEW -------------------- This dataset contains: datasheets and R Markdown files 1. Monica_groupsize_faststart.csv (datasheet used for analysis of latency) 2. Monica_groupsize_faststart no nas.csv (datasheet used for analyses of average turning rate and distance covered) 3. Monica_groupsize_NND.csv (datasheet used for analysis of nearest neighbor distance, NND) 4. Monica_groupsize_alignment.csv (datasheet used for analysis of alignment, as measured through the length of mean circular vector) 5. Kinematic performance declines as group size increases during escape responses in a schooling coral reef fish.pdf (R Markdown file in pdf format, which weaves a narrative text out of the data and associated statistical analyses in R) 6. Bacchus et al_Group size & escape response.Rmd (raw R Markdown file) Relationship between files: Files 1-4 above are the raw data associated with the statistical analyses outlined in the R Markdown files (files 5-6). -------------------------- METHODOLOGICAL INFORMATION -------------------------- Description of methods used for collection/generation of data: Full methodological details are available in Bacchus et al. 2023, https://www.frontiersin.org/articles/10.3389/frish.2023.1294259/abstract Using monofilament barrier netting, schools of juvenile blue-green Chromis damselfish C. viridis (n=336 fish) were captured from reefs in the lagoon adjacent to the Lizard Island Research Station (LIRS) in the northern Great Barrier Reef, Australia (14°40′ 08′′S; 145°27′34′′E) and immediately returned to the flow-through aquarium facilities at LIRS. Schools were maintained in groups of four, eight, and sixteen individuals each (n=36 schools; n=12 schools per treatment) in 68L tanks (65 cm L x 41 cm W x 40 cm H). Due to the possibility that differences in body size within and among schools could affect performance at both the individual and school level (Morley and Buckel, 2014), body size variation in terms of standard length was minimized both within schools (0.5 cm range from smallest to largest individual in a group) and among schools (mean standard error: 3.32  0.01 cm; range: 2.86-3.70 cm). Fish were fed a body-mass specific diet composed of freshly hatched Artemia spp. and INVE aquaculture pellets twice daily, calculated using the mean mass of all fish (1.96 g) and the number of fish in the tank (4-16 fish per tank). To ensure that each school had sufficient time to recover from the stress of collection, all fish were given 7-10 days following collection before experimental testing. This species can acclimatize to a new social context within 1-2 weeks (Nadler et al. 2021). Fish were maintained in a natural light cycle for the time of year at Lizard Island (13-hour light:11-hour dark). Experimental trials were conducted in a laminar flow swim chamber that replicated the natural flow of a coral reef on a calm-weather day (3.2 cm/s; Figure 1; Johansen, 2014). The water conditions (i.e., oxygen, temperature) in the flow chamber were maintained throughout each trial through continuous aeration and a chiller unit in the sump for the swim chamber. The working section measured 50 cm L x 40 cm W (filled to 9 cm depth), resulting in a total two-dimensional area of 2000 cm2, with each fish having 125 cm2 (in groups of 16 fish) to 500 cm2 (in groups of 4 fish) of space on average to execute behavioral responses. Food was withheld from all fish for 6-12 hours prior to testing. Once each experimental school was placed in the swim chamber, they were acclimated for four hours, as preliminary testing indicated that all shoals resumed routine swimming behavior within this time frame following introduction to the swim tunnel (Nadler et al., 2018). Following this acclimation period, the school’s escape response was stimulated using a standardized and reproducible threat protocol, in which a black tapered test tube (2.5 cm diameter × 12 cm length, 37.0 g) was released remotely from 137 cm above the arena using an electromagnet. This stimulus was discharged through a white PVC pipe (to prevent visual detection of the stimulus prior to reaching the water surface) once >50% of the school had gathered in the center of the arena (i.e., more than two body lengths from any arena wall) (Nadler et al., 2021). A piece of fishing line kept this stimulus from striking the experimental arena. Both the tapered shape of the stimulus and the fishing line aided in minimizing any ripples generated when the stimulus first made contact with the water's surface. Each trial was filmed using a high-speed video camera through a mirror that was angled 45 beneath the transparent swim chamber (240 fps; Casio Exilim HS EX-ZR1000). Between each trial, the swim chamber was drained and refilled with seawater from the LIRS flow-through system. During the experiment, we conducted two trials per day, with one in the morning and one in the afternoon, such that the acclimation for the morning trial began approximately two hours after sunrise and the afternoon trial finished approximately two hours prior to sunset. These times were balanced among group sizes. Methods for processing the data: Video recordings were examined frame by frame using the application Potplayer (v. 1.7.21566) to find crucial points in the individual's and school's response to the stimulus. Screenshots of these timepoints were analyzed in ImageJ (v. 1.53n 7). Individual escape performance was evaluated using reaction timing and kinematics, including latency (the interval between the aerial mechanical stimulus first breaking the water's surface and the fish's initial head movement), average turning rate (the maximum turning angle, , achieved by the fish during stage 1 divided by the time it took to achieve that angle, which serves as a proxy for the fish’s agility through speed of muscle contraction), and distance covered (distance moved in the first 42 ms of the reaction, which is the average time for this species to achieve stages 1 and 2 (Nadler et al., 2021); used as a proxy for swimming speed). Since these traits are influenced by the stimulus distance (the distance between the fish’s center of mass and the stimulus), this trait was also measured and included as a covariate in all analyses (Domenici and Hale, 2019). Non-responders (n=3 total, n=1 in groups of 8 and n=2 in groups of 16, 0.8% of all fish included in this study; those fish that did not respond within two seconds of the stimulus) were assigned the maximum measured latency in this study (1003.8 ms), though non-responders were not included in analyses of average turning rate and distance covered. Due to the limits that proximity to wall of the experimental arena can have on kinematic performance (Eaton and Emberley, 1991), the kinematic attributes (average turning rate and distance covered) were only assessed if the fish was >3 cm (i.e., approximately one body length) away from any wall of the experimental arena to ensure that our kinematic data was not impacted by effects imposed by the walls of the arena. Throughout the response, the school’s cohesion and coordination were measured through nearest neighbor distance and alignment, respectively. Nearest neighbor distance represents the distance between each fish’s center of mass and their most proximal neighbor’s center of mass in the school. Alignment measures the variation in each school’s members orientation with respect to the water's flow (0°) (Bachelet, 1981). This variation was quantified by using the program Oriana 4 (Kovach, 2011) and determining each school members’ angle then calculating the length of mean circular vector (r) of the group, which ranges from 0 (all members are at random angles) to 1 (all angles perfectly aligned). These characteristics were examined at intervals following the stimulus, including 0 ms (representing the school's cohesiveness and coordination immediately prior to the stimulus), 30 ms (representing the typical time for this species to complete stage 1), and 100 ms (the average time for individuals to complete both stages 1 and 2). Here, time was used as a categorical variable with these time stamps representing different stages of the escape response. R Programming Language (v. 1.3.1093) was used for all statistical analyses (R Development Core Team, 2022). The differences among treatment groups were evaluated using linear mixed-effects models (LMM), using the packages “lme4” (Walker, 2015), “car” (Weisberg, 2019), “MuMIn” (Bartoń, 2023), “emmeans” (Lenth, 2022) and “ggplot2” (Wickham, 2016). Latency, average turning rate, and distance covered were analyzed using LMMs, with group size as a fixed effect, stimulus distance as a covariate, the interaction between group size and stimulus distance, and school identifier as a random effect (so that each individual was nested within their school). School traits (nearest neighbor distance and alignment) were measured with group size, time post-stimulus (0, 30, 100 ms), and their interaction as fixed effects. For the alignment analysis, school identifier was included as a random effect to account for the repeated measures design. Nearest neighbor distance was measured on an individual level, so individual was nested within the school identifier as a random effect. To verify that the assumptions of normality and homogeneity of variance were met for each model, we visually inspected the quantile-quantile and residuals plots and used Shapiro-Wilk and Bartlett tests. To meet these assumptions, all response variables (except average turning rate and alignment) were boxcox transformed using the package "car" (Weisberg, 2019). Best-fit models were identified using the Akaike Information Criterion (AIC; Hunsicker et al., 2011) model selection. When significant fixed effects were identified, they were further explored using Tukey’s multiple comparison post hoc tests. The R2 for all models is also detailed below, including the marginal and conditional R2 (R2m and R2c, respectively), which represent the variance explained by the fixed effects only (R2m) and the fixed effects plus the random effects (R2c). Software used to interpret the data: Potplayer (v. 1.7.21566), ImageJ (v. 1.53n 7), R Programming Language (v. 1.3.1093), Oriana 4 Quality-assurance procedures performed on the data: All video analyses and entered data were double-checked by a second person. When results differed among two observers, and could not be resolved, a third observer conducted the analyses to determine the correct values. People involved with sample collection, processing, analysis and/or submission: Fish Collection: Lauren Nadler, Eva McClure, Mark McCormick Escape Response Trials: Lauren Nadler, Eva McClure, Shaun Killen Video Analyses: Lauren Nadler, Monica Bacchus, Robby Spekis, Alianna Jones, Delaney Farrell, James Puentes, and Raghavi Vuppala Statistical Analyses: Lauren Nadler, Monica Bacchus -------------------------- DATA-SPECIFIC INFORMATION -------------------------- 1. Monica_groupsize_faststart.csv Number of variables: 10 Number of cases/rows: 337 Variable list, defining any abbreviations, units of measure, codes or symbols used: A. Group size: number of fish in the group (4, 8, or 16) B. Video: Denotes the video code (G denotes group size, G4/G8/G16, followed by the number of that video trial, 1-12). C. Fish: Refers to that fish's number in the group (starting at 1 and going up to 4 in groups of 4, up to 8 in groups of 8, up to 16 in groups of 16). D. Fish reacts?: Denotes if the fish reacts with 2 seconds of the stimulus (y = yes, n = n). E. Latency: The interval between the aerial mechanical stimulus first breaking the water's surface and the fish's initial head movement, measured in the unit millisecond (ms). F. Responder number: Denotes the ordering of responder number in the group, using standard competition ranking. This variable was not used in the final analyses. G. >1BL (3 cm): Specified if the fish was greater than 1 body length (BL) from any wall of the experimental arena (3cm; y = yes, n = no). H. Average turning rate: the maximum turning angle achieved by the fish during stage 1 divided by the time it took to achieve that angle, which serves as a proxy for the fish’s agility through speed of muscle contraction. Measured in the unit degrees per millisecond. I. Distance covered: Distance moved in the first 42 ms of the reaction, which is the average time for this species to achieve stages 1 and 2, which is used as a proxy for swimming speed. Measured in the unit cm. J. Stimulus distance: The distance between the fish’s center of mass and the stimulus, measured in cm. Missing data codes: Empty cells under the average turning rate and distance covered columns occurred when any fish was <1 body length (BL, 3 cm) from any wall of the experimental arena (as denoted in the column labelled ">1BL (3cm)"; y = yes the fish was >3cm from any wall, n = no the fish was <3cm from a wall). These cells could also be empty if the fish didn't respond at all (as denoted "n" in the "Fish reacts?" column and NR in the ">1BL (3cm)"). 2. Monica_groupsize_faststart no nas.csv Number of variables: 10 Number of cases/rows: 337 Variable list, defining any abbreviations, units of measure, codes or symbols used: A. Group size: number of fish in the group (4, 8, or 16) B. Video: Denotes the video code (G denotes group size, G4/G8/G16, followed by the number of that video trial, 1-12). C. Fish: Refers to that fish's number in the group (starting at 1 and going up to 4 in groups of 4, up to 8 in groups of 8, up to 16 in groups of 16). D. Fish reacts?: Denotes if the fish reacts with 2 seconds of the stimulus (y = yes, n = n). E. Latency: The interval between the aerial mechanical stimulus first breaking the water's surface and the fish's initial head movement, measured in the unit millisecond (ms). F. Responder number: Denotes the ordering of responder number in the group, using standard competition ranking. This variable was not used in the final analyses. G. >1BL (3 cm): Specified if the fish was greater than 1 body length (BL) from any wall of the experimental arena (3cm; y = yes, n = no). H. Average turning rate: the maximum turning angle achieved by the fish during stage 1 divided by the time it took to achieve that angle, which serves as a proxy for the fish’s agility through speed of muscle contraction. Measured in the unit degrees per millisecond. I. Distance covered: Distance moved in the first 42 ms of the reaction, which is the average time for this species to achieve stages 1 and 2, which is used as a proxy for swimming speed. Measured in the unit cm. J. Stimulus distance: The distance between the fish’s center of mass and the stimulus, measured in cm. *This spreadsheet is similar to the previous file, but has the blank cells removed as they caused errors when conducting the statistical analysis in R. 3. Monica_groupsize_NND.csv Number of variables: 5 Number of cases/rows: 1009 Variable list, defining any abbreviations, units of measure, codes or symbols used: A. Groupsize: number of fish in the group (4, 8, or 16) B. Video: Denotes the video code (G denotes group size, G4/G8/G16, followed by the number of that video trial, 1-12). C. Fish: Refers to that fish's number in the group (starting at 1 and going up to 4 in groups of 4, up to 8 in groups of 8, up to 16 in groups of 16). D. Time: School traits (nearest neighbor distance and alignment) were measured at intervals following the stimulus, including 0 ms (representing the school's cohesiveness and coordination immediately prior to the stimulus), 30 ms (representing the typical time for this species to complete stage 1), and 100 ms (the average time for individuals to complete both stages 1 and 2). Here, time was used as a categorical variable with these time stamps representing different stages of the escape response. E. NND: Nearest neighbor distance (NND) represents the distance between each fish’s center of mass and their most proximal neighbor’s center of mass in the school, measured in cm. 4. Monica_groupsize_alignment.csv Number of variables: 4 Number of cases/rows: 109 Variable list, defining any abbreviations, units of measure, codes or symbols used: A. Video: Denotes the video code (G denotes group size, G4/G8/G16, followed by the number of that video trial, 1-12). B. Groupsize: number of fish in the group (4, 8, or 16) C. Time: School traits (nearest neighbor distance and alignment) were measured at intervals following the stimulus, including 0 ms (representing the school's cohesiveness and coordination immediately prior to the stimulus), 30 ms (representing the typical time for this species to complete stage 1), and 100 ms (the average time for individuals to complete both stages 1 and 2). Here, time was used as a categorical variable with these time stamps representing different stages of the escape response. D. Alignment: Measures the variation in each school’s members orientation with respect to the water's flow (0°). This variation was quantified by using the program Oriana 4 and determining each school members’ angle then calculating the length of mean circular vector (r) of the group, which ranges from 0 (all members are at random angles) to 1 (all angles perfectly aligned).