Dataset: Phytoplankton Orientation - Diatom Stephanopyxis

A uni-algal culture of the colonial diatom Stephanopyxis turris was grown in 2 separate Couette chambers under controlled shear conditions for up to 24 days. The Couette chambers (40 cm height) consisted of an inner rotating cylinder (12.7 cm radius) and an outer stationary cylinder (14.92 cm radius). A laminar flow chamber rotated at 0.5 rpm and a turbulent flow chamber rotated at 13 rpm. The gap between the cylinders was filled with 8 L of sterile L1 medium (Guillard and Hargraves 1993). Experiments took place in a temperature and light controlled walk-in incubator at 20° C with a 12:12 light:dark cycle. Light was provided by cool white fluorescent lamps producing 32 µmol photons m-2 s-1 at the center of each chamber.

Digital holograms through the entire height of each Couette chamber (40 cm) were acquired at 3 day intervals during the duration of the experiment. Each hologram imaged 19.35 mL of water. The total volume imaged and analyzed on each day and for each chamber was 3.9 L representing 202 holograms.

A custom digital holographic microscope (DHM) was constructed to image particles within the Couette chambers. Coherent illumination was provided by a green (532 nm) nanosecond pulsed laser. The incident beam was directed upward through the bottom of the chamber and images were acquired by a CCD camera (4896 x 3264 pixels) positioned above the chamber looking downward through the illuminated volume. An objective lens in front of the camera increased magnification and positioned the hologram image plane at the water surface. Hologram resolution was 1.74 µm/pixel.

Holograms were numerically reconstructed in Matlab (version 2019a) using the Kirchoff-Fresnel convolution kernel (Katz and Sheng 2010) at 1 mm intervals throughout the height of the chamber. Reconstructed images were combined to produce a single extended depth of field (EDF) image for each hologram. These EDF images were segmented with a fixed threshold and particle measurements were obtained through automated region analysis in Matlab. See Nayak et al. 2018 for further details.