March 2nd 2015 This is the ReadMe file for the data accompanying the 2015 van der Bijl et al. paper on the effect of brain size on the predator response in guppies. For further questions mail to wouter.van.der.bijl@zoologi.su.se or woutervdbijl@hotmail.com. Brief overview of experiment: After a night of habituation we exposed guppies to a predator model and a novel object control. Either as single fish, pairs or shoals of four. All groups were single sex (male or female), of a single brain size (either small or large) and from a single replicate selection line (1, 2 or 3). Each group did both a predator and control trial. There are two data sources used: 1. An observer manually scored predator inspection behavior from video, 2. Automatic video tracking software (ctrax) was used to extract locations and orientations of the fish. ####################### ### Inspection data ### ####################### **Raw** data is available as JWatcher output in 'Raw_inspection_data.rar'. This file contains a folder for each day, and a .dat file for each trial. .dat files are the format of JWatcher and can be opened in a text editor. First lines contain meta data. The actual data table indicates keypresses with their times in miliseconds. A guide for the meaning of the keypresses is available in the file 'Coding legend.txt' in the root directory. **Processed** data is available in 'Processed_inspection_data.csv'. All times are in miliseconds. An overview of the not-obvious variables: - treat c is control trial, p is predator trial - model c is control mug, old and new denote two seperate predator models - exp.tank The tank the experiment was performed in, exp.tank nested in day gives unique id's for each shoal. - bs Brain size, 0 is small 1 is large - sex 0 is female, 1 is male - rep Selection line replicate - group Shoal size, 1 2 or 4 - poke Number of times a fish poked the model - burst Number of observed bursts/skitters Slide variables measure the side swimming behavior, while inspections are close bouts of slides (they are very strongly correlated, and we only used inspections). 1234 denote group size during inspection, for example: - tot.insp.duration total duration of all inspections - 1.tot.insp.duration total duration in miliseconds of inspections of a single fish - 4.tot.insp.duration total duration of inspections when four fish inspected simulatiously Other variables: - tot.insp.count number of inspections performed - avg.insp.duration average duration of inspection - f.insp.duration total duration of inspection where inspection group is taken into account, so 1 second of 4 fish inspecting counts as 4. - insp.joins number of times where the inspection group size increased during inspecting - insp.abandons number of times where the inspection group size decreased during inspecting ##################################### ### Location and orientation data ### ##################################### Note: while the tracking data has individual identities assigned, there is NO guarantee that identities are not frequently switched during the trial. In other words, id 1 may represent different fish at different times. **Raw** data is available as the corrected output of ctrax in 'Raw_tracks_control_trails.rar' and 'Raw_tracks_predator_trials.rar'. Each trial has one .csv named as 'fixed_d[day]t[trial].csv'. The meta-data about what sex, brain size, replicate etc. this trial represents can be found in 'Processed_inspection_data.csv' using the day and trial variables (see above). The .csv files follow the ctrax format (http://ctrax.sourceforge.net/ctrax.html#file-export-as-csv-file). The table is a nframes x 7*nfish rectangular matrix. Each row is a single frame (30fps) and each group of 6 consecutive colums belongs to one fish. The columns (in groups of six) have the following meanings: 1. id (-1 means the fish was not tracked in that frame) 2. x coordinate (in pixels) 3. y coordinate (in pixels) 4. length of the fit ellipse (in pixels) 5. width of the fit ellipse (in pixels) 6. orientation (radians) Additionally, in 'Video_snapshots.rar' there a screengrabs from each of the trials that can be used to extract tank and object coordinates. **Processed** data is available in 'Processed_tracks.rar'. Each trial is named 'd[day]t[trial].csv'. The meta-data about what sex, brain size, replicate etc. this trial represents can be found in 'Processed_inspection_data.csv' using the day and trial variables (see above). These tracks have undergone the following steps of post-processing: 1. Some sanity checks excluding (very few) frames, such as tracked locations outside the tank. 2. For all but two trials the y-axis and orientations needed to be flipped (idiosyncrasy of ctrax). 3. An image transform was used to fit the coordinates to a rectangular plane (see paper). Orientations needed correcting to go along with this. 4. Using translation and rotation the head of the predator or front of the cup is defined as (0, 0) and the x-axis is oriented along the body of the predator (with the tail towards negative). This facilitates comparison between trials. 5. Areas with tracking problems have been identified. The .csv files are in long format with the following variables: - fish Fish id (not guaranteed to be maintained over time) - frame Time indication (30 frames per second) - x x location in cm - y y location in cm - orientation Orientation in radians - not.exact Reliability information. NA indicates no problems, 1 indicates inaccuracy at the left wall, 2 indicates inaccuracy at the object (predator model or coffee cup).