Individual zebrafish larvae (n = 18) and guppy neonates (n = 26) were isolated within a small (3.57 ×10⁻⁶ m³) custom‐built plastic tank (using optically clear plastic from Edmund Optics Inc. Barrington, NJ), with inner dimensions: 0.00703 m × 0.02022 m × 0.02512 m (depth × width × height). Fish were acclimated for 2 min prior to filming. Video sequences were captured using a Photron FASTCAM MiniAX (Photron USA, Inc., San Diego, CA) fitted with an Infinity Photo‐Optical Company lens (Boulder, CO). The filming area was illuminated using fiber‐optic illumination (ThorLabs OSL2, Newton, NJ). Following the period of acclimation, ~25 individual prey items were delivered to the 3.57 ml volume feeding chamber. In an effort to match relative prey and predator size, we used Paramecium (100–300 μm length) for filming zebrafish, and Artemia nauplii (400–500 μm length) for filming guppies. All hydrodynamic manipulations made to the experimental treatments equally affected both the fish and their prey (i.e., the effect of viscosity manipulation was linear).

Three sets of video recordings were made to gather three separate sets of data: (a) strike and capture rate data were obtained from video at 250 fps, which recorded continuously for 48 seconds, over which feeding performance was measured; (b) data for kinematic analyses were filmed at high‐speed (1,000 frames per second; fps) to measure feeding kinematics; and (c) flow visualization data were obtained with video recordings at 1000 fps with suspended neutrally buoyant particles to obtain PIV, which we used to determine flow fields during in vivo and unrestrained prey capture.

Following each feeding trial, fish were euthanized by an overdose of tricane methanosulfonate (Tricane‐S, Western Chemical, Inc., Ferndale, WA) buffered in sodium bicarbonate (Fisher Scientific, Fair Lawn, NJ). Specimens were fixed in 4% buffered paraformaldehyde (Sigma, St. Louis, MO) overnight and transferred to 70% ethanol for long‐term storage.