Surprisingly, the fastest motions are not produced by large animals or robots. Rather, small organisms or structures, including cnidarian stinging cells, fungal shooting spores, and mandible strikes of ants, termites, and spiders hold the world acceleration records. These diverse systems share common features: they rapidly convert potential energy - stored in deformed material or fluid - into kinetic energy when a latch is released. However, the fastest and smallest known movements often cannot be used multiple times, because mechanical components are broken or ejected. Furthermore, some of these systems must overcome the added challenge of moving in water, where high density and viscosity constrain acceleration at small sizes. Here we report the kinematics of repeatable, ultrafast snaps by tiny marine amphipods (Dulichiella cf. appendiculata). Males use their enlarged major claw, which exceeds 30% of body mass, to snap a 1 mm-long dactyl with a diameter equivalent to a human hair (184 µm). The claw snaps closed extremely rapidly, averaging 93 µs, 17 m s-1 and 2.4 x 105 m s-2. These snaps are among the smallest and fastest-moving of any documented repeatable movement and are sufficiently fast to operate in the inertial hydrodynamic regime (Re >10,000). They generate audible pops and rapid water jets, which occasionally yield cavitation, and may be used for defense. These amphipod snaps push the boundaries of acceleration and size for repeatable movements, particularly in water, and exemplify how new biomechanical insights can arise from unassuming animals.

Videos for kinematic analyses were recorded at 300,000 frames s⁻¹ (256x128 pixel resolution, 2.33 μs shutter duration, FASTCAM SA-Z type 2100K-M-64GB, Photron, San Diego, CA, USA). 

See Usage notes below and the Current Biology online supplement for our detailed methods. 

General notes: High speed video files are saved in .avi format. Their names begin with the video collection date, followed by “_Amphipod_” then the individual ID (the letter A or B with a 1 or 2 digit number), followed by another underscore and the video number for that individual. Some file names have additional letters and numbers following this that are not pertinent to this study. For instance, “071018_Amphipod_A2_4_DAQ12” refers to the fourth video collected for individual A2 on July 10, 2018. Steps throughout our analyses preserve Individual ID and video numbers. 

Dryad submission contents:

1.    Snap_Example_071218_Amphipod_A11_2.avi: Example video of a gnathopod snap. This video was filmed at 300,000 fps. For our analysis, the strike parameters were quantified from frame 6 to frame 30. This video has been cropped and bit shifted to aid in visualization. 

2.    Cavitation_Example_071118_Amphipod_A5_1.avi: Example video of a gnathopod snap that generates a cavitation bubble. The bubble is first visible on frame 25 and completes initial collapse on frame 35, followed by addition bubble rebound. This video was filmed at 300,000 fps and plays back at 30fps. This video has been cropped and bit shifted to aid in visualization

3.    Amphipod_kinematics_ALL_FINAL_wIDs.csv contains the kinematics for all 60 strikes in the dataset. Details about each column heading are as follows: ID_supplemental_table1: The first column “ID_supplemental_table1” has the ID number (1-16) arbitrarily given to each individual in Supplemental Table 1 in the Current Biology online supplement; ID: Individual specimen identifier; Video: Name of each individual video file. Usually the date collected [underscore] Amphipod [underscore] ID [underscore] video number; bitShift: indicate the value by which the video sequence was bit-shifted prior to digitization (a value of zero indicates that there was no bit-shift applied); firstFrame: number of frame that marks the beginning of the strike; endFrame: number of the frame that marks the end of the strike; dactyl_px: dactyl length in pixels determined from the video sequence; propodus_px: propodus length in pixels determined from the video sequence; num_strikes_analyzed: the number of strikes in total from that individual amphipod (including the current entry);    Date_vid_collected: date that the video sequence was collected;    Scale_by: indicates whether dactyl length or propodus height as measured under the microscope was used to scale the video sequence; Dactyl_length_for_scale_mm: if applicable, the dactyl length in millimeters measured from scope images of the individual to be used to scale the video sequence; Propodus_length_for_scale_mm: if applicable, the propodus height in millimeters measured from scope images of the individual to be used to scale the video sequence; mean_total_dry_mass_mg: the mean total dry mass of the amphipod reported in milligrams. A mean is reported because three measurements were made and averaged;    mean_claw_dry_mass_mg: the mean total dry mass of dactyl and propodus reported in milligrams. A mean is reported because three measurements were made and averaged; mean_propodus_dry_mass_mg: the mean total dry mass of propodus after removal of the dactyl is reported in milligrams. A mean is reported because three measurements were made and averaged; mean_dactyl_dry_mass_mg: the mean total dry mass of dactyl is reported in milligrams. This value was calculated as the difference in the mean total claw mass and mean propodus mass for that individual; vid_dactyl_length_mm: the length of the dactyl reported in millimeters for that individual. This is either the length of the dactyl measured directly on the individual under a microscope and reported as "Dactyl_length_for_scale_mm" or it is calculated from "Propodus_length_for_scale_mm" using the lengths of the propodus and dactyl in pixels; calculated_vidScale_ppmm: the scale used to calibrate the video in pixels per millimeter (ppmm). This scale is determined using the known dimensions of the dactyl or propodus from the microscope and their lengths in pixels measured from the video; duration_frame: the duration of the strike in number of frames; duration_s: the duration of the strike in seconds; angle_t1_deg: the angle between the line on the dactyl and the line on the propodus at the start of the strike in degrees (see full methods in paper for details); angle_t2_deg: the angle between the line on the dactyl and the line on the propodus at the end of the strike in degrees (see full methods in paper for details); rotation_deg: change in angle in degrees between the start and end of the strike; rotation_rad: change in angle in radians between the start and end of the strike; average_rot_v_dps: the average rotational velocity over the course of the strike in degrees per second (dps); average_rot_a_dps2: the average rotational acceleration over the course of the strike in degrees per second squared (dps2); average_rot_v_rps: the average rotational velocity over the course of the strike in radians per second (rps); average_rot_a_rps2: the average rotational acceleration over the course of the strike in radians per second squared (rps2); average_speed_mmps: the average linear speed over the course of the strike in millimeters per second; average_speed_mps: the average linear speed over the course of the strike in meters per second; average_at_mmps2: the average linear acceleration over the course of the strike in millimeters per second; average_at_mps2: the average linear acceleration over the course of the strike in meters per second. 

4.    Digitize_amphipod_videos_inFiji_FINAL.ijm is a macro script for Fiji that was used to automate the digitization of videos for our analyses. With a video open, run the script. It will prompt you to choose a location for the output and then walk you through the process to digitize the first and last frame of the strike, as well as to measure the scale, dactyl and propodus lengths. 

5.    final_dig_files.zip contains the digitizing output for all 60 strikes in our dataset. Additionally, the output subfolder contains output from the kinematic analyses performed in R (see below). There are 60 files pertaining to the kinematic output for each video in the dataset (Amphipod_kinematics_[video name]_final.csv), the master output file for all 60 videos (Amphipod_kinematics_ALL_FINAL.csv) and a clean summary of kinematics by individual used to create the supplemental table (Amphipod_kinematics_SUPPLEMENTAL_TABLE_FINAL.csv). Information about column headers can be found in the file "Digitization Files READ ME.txt".

6.    metadata.zip contains metadata for individuals (e.g., masses, dactyl lengths) and videos (e.g.,  date, framerate, name of the person who digitized, etc). Individual summary data can be found in : Amphipod_Individual_SUMMARY_dryad.csv and video data can be found in Amphipod_Diglog_dryad.csv. These files serve as additional inputs in the R script below. Information about column headers can be found in the file "Metadata READ ME.txt".

 
7.    Amphipod_Kinematics_from_Digitizing_Rscript_FINAL.R computes and summarizes kinematics using the digitization files and metadata above as inputs. 

8.    Comparative_dataset_FINAL.csv: Dataset used to create the comparative plot in Figure 1. See supplemental methods for more details. Details about each column heading are as follows: PLOTCODE: code used to facilitate plotting in R; MASSKG: mass of actuated structure in kilogram; MASS_g: mass of actuated structure in grams; ACCELMpSsqd: acceleration in meters per second squared; Citation: citation for papers where information for mass and acceleration were taken from or derived.