Surface (1 m) water samples were collected using precleaned 10 L HDPE canisters (Plastex®, Finland) and then filtered through 0.2 μm Whatman Polycap 36 AS nylon membrane cartridge filters, using a Masterflex L/S Digital Standard Drive peristaltic pump with Easy Load II and Masterflex L/S 17 silicone tubing at 100 mL min−1, directly into precleaned 10 L polyethylene Hedwin CubitainersTM. Filtered sample waters were stored in the dark in an environmental chamber set at temperatures matching the in situ temperatures before use.
Samples from each treatment group were partitioned into six 10-cm-pathlength cylindrical Spectrocell spectrophotometric quartz cells (~30 mL volume each). Each cell was rinsed three times with sample water from the corresponding cubitainers, and then filled without headspace directly from the cubitainers. Each cell was capped with two Spectrocell caps fitted with Microsolv Teflon-lined butyl septa. Five cells for each treatment type (15 total) were placed vertically into a temperature (15°C)-controlled black aluminum block, and irradiated under an Atlas Suntest CPS+ solar simulator equipped with a 1.5 kW xenon lamp. The solar simulator was fitted with a daylight filter (excluding light below ~300 nm) to provide the cells with precisely known, full spectral light. One cell for each treatment type was wrapped in aluminum foil to serve as dark control and placed in the same water bath that provided cooling water to the aluminum block. Absorbance (A(λ)) in the control and irradiated samples were measured at 250–800 nm at 1.0 nm intervals, in duplicate, in a 1-cm-pathlength quartz spectrophotometric cuvette. At the Tvärminne Zoological Station, a Shimadzu UV-2501PC UV-VIS recording spectrophotometer with UV-Probe software was used, with air as an internal reference and Milli-Q water as blanks.
At Skidaway Institute of Oceanography, absorbances were measured using an Agilent 8453 UV-visible spectrophotometer with ChemStation software, with the same parameters as above and Milli-Q water as blanks. The average absorbance spectra of blanks were subtracted from the absorbance spectra of the sample water. Absorbance spectra were further corrected for potential offsets and instrument drift by subtracting the average absorbance at 690–710 nm, before converting A(λ) to Napierian absorption coefficients (ag(λ); m−1), using the following equation: ag(λ)= A(λ) ln10/L, where L (m) is the pathlength.