Copepods were incubated with phytoplankton for 1 hour. Incubations were terminated by gently pouring the contents of incubation bottles through a 53 mm sieve, then rinsing the copepods into a petri dish with GF/F filtered water. A total of 12 copepods were sampled from each stage for each phytoplankton and for the controls.

Copepods were sorted under a dissecting microscope and immediately transferred to glass microscope slides. Excess water was removed with a fine-tipped Pasteur pipette to immobilize the copepod for optimal viewing. To minimize pigment degradation, slides containing the copepods were transferred to the microscope as quickly as possible, always within 2 min of sorting.

A Nikon model E400 epifluorescence microscope at 20X magnification, outfitted with a Nikon B-2A long-pass filter cube (470EX/515LP, Nikon) and a custom band-pass filter cube (430EX/680EM, Omega Optical) was used. The longpass filter pair provided a brighter overall image than the band-pass filter pair and also fluoresced non-chlorophyll structures (e.g., the copepod exoskeleton fluoresces green with blue excitation). The band-pass filter cube showed fluorescence from only chlorophyll and phaeopigments. Photographs of the copepods were taken at each filter setting using a Canon Digital model T3i single-lens reflex camera (ISO 6400, F 1/15) remotely controlled with the Canon Electro-Optical System Utility software to produce clear images of the epifluorescence (Vogt, 2013,Fig. 1).

Images were processed with Photoshop CS6 (Adobe). Total gut area was manually digitized using the lasso and measurement tools on the long-pass-filtered images. The mean gut area of 12 individuals for each copepod species and stage was used for calculations. The threshold tool was used to select areas of the long-pass-filtered images containing pigment (gut pigment area) that exceeded baseline fluorescence. Signal intensity was not used in the calculation of the GPI (Gut Pigment Index) because pigment content and composition differed among phytoplankton taxa. Pixels of the band-pass-filtered images exceeding the threshold value were selected and their area estimated using the measurement tool. The gut-pigment area was then divided by the total gut area, yielding a relative GPI for each copepod. The same procedure was used to process images of copepods sampled from the control bottles. Copepods with <5% GPI were assumed not to have fed, as the GPI values in those samples were within the range of indices estimated from the controls.