Physiological and interspecific factors determine diel changes in phytoplankton bio‐optical properties

Bio‐optical properties of marine phytoplankton retrieved through satellite remote sensing are used to estimate ocean productivity and carbon cycling. Daily activity of phytoplankton is attuned to the predictable light fluctuations of the diel cycle. Field and laboratory studies have documented diel changes in phytoplankton growth, division, and respiration, carbon and pigment content, cell size, photosynthetic efficiency and rate, and DNA replication and transcription. Many of these physiological changes can alter cellular optical properties and contribute to diel variations in bulk absorption and scattering properties. Consequently, understanding phytoplankton contributions to diel optical cycles is essential for improving algorithms that convert remote sensing data to biological rates and stocks. Here, we describe time‐resolved cellular, photophysiological, and bio‐optical properties for three cultured phytoplankton ranging in cell size from ~ 1 to 6 μm: Ostreococcus lucimarinus, Synechococcus (WH8102), and Thalassiosira pseudonana, all of which can significantly contribute to phytoplankton abundance and/or biomass in the open and coastal ocean. This work is the first to characterize complete diel cycles in absorption and attenuation for O. lucimarinus and backscattering for all these species. Results show that the percent increase from the minimum to maximum values over the diel cycle ranged between ~ 24–121%, ~ 31–235%, and ~ 25–198% for particulate absorption, attenuation, and backscattering, respectively. Diel changes in bio‐optical properties also differed in both timing and magnitude across phytoplankton species, demonstrating the importance of contextualizing remote sensing observations with phytoplankton community composition.

The submarine light environment is influenced by properties of resident phytoplankton populations, including cell concentration, size distribution, cell shapes, pigment composition, internal cellular structures, and cell refractive indices (Kirk 1975;Jerlov 1976).Attenuation, absorption, and scattering coefficients of phytoplankton are additive inherent optical properties (independent of the angular distribution and magnitude of the incident light field) needed in forward bio-optical models to interpret apparent optical properties (dependent on the angular distribution and magnitude of the light field) of seawater observed by satellite remote sensing (Morel and Prieur 1977;Bricaud et al. 1983).Knowing the contribution of phytoplankton absorption and scattering allows satellite measurements of water-leaving radiance to be used in inverse bio-optical models to monitor changes in phytoplankton biomass and production, which are critical for forecasting ocean food web and biogeochemical processes (Morel et al. 1991;Behrenfeld et al. 2005;Zaneveld et al. 2007).
The daily fluctuation of light in the surface ocean is an important environmental cue for phytoplankton.Many phytoplankton synchronize endogenous biological clocks with the daily light cycle to regulate and prevent induction of nightassociated metabolic processes during ephemeral exposures to darkness (e.g., cloud cover, storm-driven deep mixing) (Sournia 1975;Brand 1982).Existence of such endogenous clocks is demonstrated in laboratory phytoplankton cultures exhibiting periodicity in photosynthesis for > 70 h following transition from light to dark cycles to constant light (e.g., Brand 1982).
Diel physiological cycles drive corresponding bio-optical cycles.For example, cell and/or carbon-specific attenuation coefficients are influenced more by changes in cellular refractive index than cell size for the diatom Thalassiosira pseudonana (Stramski and Reynolds 1993) and cyanobacterium Synechococcus (WH8103) (Stramski et al. 1995), while the opposite was observed for two small chlorophytes, Nannochloris and Micormonas pusilla (DuRand and Olson 1998;DuRand et al. 2002).Thus, understanding such species-specific dependences is essential for predicting bulk seawater optical properties from phytoplankton community composition (Ahn et al. 1992) and vice versa (Siegel et al. 1989;Stramska and Dickey 1992).
Diel cycles in bio-optical properties, namely absorption, attenuation, and scattering, have been observed for a range of cultured phytoplankton (Stramski and Reynolds 1993;Stramski et al. 1995;DuRand and Olson 1998;DuRand et al. 2002;Poulin et al. 2018).Building on this foundation, the current study reports complete diel cycles in backscattering coefficients alongside other cellular, physiological, and optical properties for T. pseudonana, Synechococcus (WH8102), and the chlorophyte Ostreococcus lucimarinus.These species were chosen to represent phytoplankton groups that are often numerically dominant in the ocean.Diatoms (e.g., T. pseudonana) often dominate nanophytoplankton (5-20 μm) biomass and abundance in upwelling and high latitude regions (Pierella Karlusich et al. 2020).Synechococcus and Ostreococcus often contribute significantly to picophytoplankton (≤ 2 μm) abundance and biomass in coastal and open ocean waters (Worden et al. 2004;Bolaños et al. 2020).
Cell-and carbon-specific changes in absorption, attenuation, and scattering have been observed to be related to changes in cellular refractive index and cell size (Stramski and Reynolds 1993;Stramski et al. 1995).Poulin et al. (2018) used T. pseudonana cultures to describe cell size-and abundance-dependent diurnal changes in backscattering.
Here, we report the first complete diel cycles in bio-optical properties for O. lucimarinus and backscattering coefficients for T. pseudonana and Synechococcus.We show that (1) biooptical properties co-vary with diel changes in cell physiology, (2) interspecific variability exceeds intraspecific variability in diel cycles of biomass-specific bio-optical properties, and (3) diel variations of biomass-specific beam attenuation and backscattering coefficients differ in amplitude and timing within and between species.These results are particularly relevant for understanding day-night differences in ocean optical properties observed with satellite lidar (e.g., Cloud-Aerosol Lidar with Orthogonal Polarization [CALIOP] sensor; Behrenfeld et al. 2019), but are also important for interpreting ocean color data from geostationary sensors and between polar-orbiting sensors with different equator crossing times.

Culture growth conditions
F/2 + Si medium was used to grow T. pseudonana (Guillard and Ryther 1962), L1 medium to grow Synechococcus (Guillard and Hargraves 1993), and L1-Si medium to grow O. lucimarinus.All phytoplankton were maintained in semicontinuous growth by dilution every 1-2 d for at least 3 weeks to ensure cells were fully acclimated to their environment.During the weeks prior to the full suite of bio-optical sampling, cell concentrations were monitored once or twice daily at the same times of day and variable fluorescence was monitored to ensure consistent growth and physiology during diel sampling.Cultures were maintained in 1-liter polycarbonate bottles in a temperature-controlled room at 18 C under a 12 : 12-h light-dark sinusoidal light cycle.Light was provided using a custom light-emitting diode (LED) Phyto-panel (Photon Systems Instruments) composed of white and blue light LEDs.Irradiance ranged from total darkness to 390 μE m À2 s À1 and was measured with a 4π spherical quantum meter (Biospherical Instruments QSL-100) (Fig. 1a).All cultures were continuously bubbled with ambient air filtered through an activated charcoal hydrocarbon trap to keep cells in suspension and to prevent CO 2 limitation.

Diel experiments
Two independent experiments were conducted for each phytoplankton species within a week of each other and are summarized in Table 1.During each experiment, acclimated cultures were sampled every 2 h over the course of 24-28 h.While the total number of replicate cultures and sampled parameters varied between experiments, samples were always taken for flow cytometry and chlorophyll fluorescence to evaluate consistency between replicate cultures and experiments.In both T. pseudonana experiments, cultures were diluted to dawn cell concentrations at each measured time point after the onset of cell division 8 h after dawn to ensure sufficient sampling volume for the duration of the experiments.Synechococcus and O. lucimarinus experiments were not diluted during the experiments as these species were maintained at cell densities 1-2 orders of magnitude higher than T. pseudonana and thus required less sample volume per sampling time point.Flow cytometry measurements Samples of live phytoplankton cells were collected directly from cultures and from the recirculating optical sampling system following measurement of attenuation, absorption, and backscattering.These samples were immediately enumerated and analyzed using a Guava easyCyte 5HT flow cytometer (Millipore) with a 50-mW blue laser (488 nm), three bandpass filters (525 AE 30, 583 AE 26, and 695 AE 50 nm), and detectors for forward and side scattering (FSC and SSC, respectively).T. pseudonana and O. lucimarinus cell fluorescence was detected at 695 nm (chlorophyll fluorescence indicator), and Synechococcus fluorescence was detected at 583 nm (phycoerythrin fluorescence indicator).

Carbon and nitrogen
Particulate organic carbon (POC) and particulate organic nitrogen (PON) concentrations (mg C and N m À3 ) were determined in triplicate from 5 to 6 mL culture volumes (5.4 AE 0.3 mL) filtered onto pre-combusted (450 C for 5 h) 25 mm Advantec GF75 filters (0.3 μm, nominal).Due to their small size, Ostreococcus were filtered through two stacked filters to increase filter cell retention.Filtrates were analyzed by flow cytometry to assess cell loss through filters.In filtrate samples where cells were detected (n = 29 of 191 samples), the total number of cells in the filtrate represented 1.1% AE 0.9% of the unfiltered cell count.Culture filtrates were also re-filtered onto GF75 filters to be analyzed for extracellular POC and PON background subtraction.Filters were immediately placed in storage at À20 C and subsequently dried at 60 C for 24 h, packed into tin capsules, and stored in a desiccator until analysis.Elemental composition was determined with an Exeter Analytical CE-440 Elemental Analyzer calibrated using acetanilide standards.Calculated POC and PON for each sample were normalized to the respective cell concentration (cells m À3 ) and were also used to assess cellular carbon to nitrogen (C : N, mol : mol) and carbon to chlorophyll a (C : Chl a, weight : weight) ratios.

Chlorophyll fluorescence
Measurements of chlorophyll fluorescence yield when all functional PSII reaction centers are oxidized (F o ) and reduced (F m ) were made for each culture using a Soliense Light Induced Fluorescence Transient Fluorometer (LIFT-FRR) (Kolber et al. 1998).Synechococcus and O. lucimarinus samples were dark-acclimated for 5 min to relax non-photochemical quenching (NPQ) (Mackey et al. 2013;Halsey et al. 2014).T. pseudonana samples were low-light acclimated at 10-15 μEin m À2 s À1 prior to measurements because diatom NPQ does not fully relax in the dark (Milligan et al. 2012).For each sample, 200-300 μL of culture volume were diluted into 2 mL of 0.2-μm filtered artificial seawater and then measured in a glass cuvette within the LIFT-FRR.We followed the four-phase FRR excitation protocol described by Brown et al. (2019) and used data from the initial single-turnover phase to calculate the quantum efficiency of photochemistry (F v /F m ) as (Maxwell and Johnson 2000): Optical measurements Hyperspectral absorption (a) and attenuation (c) of the cultures and 0.2-μm filtered culture water were measured at $ 4 nm spectral resolution using an AC-S spectrophotometer (Sea-Bird Scientific [WET Labs, Inc.]; serial number 94).Simultaneous backscattering counts at three wavelengths (470, 532, and 660 nm) were measured using an ECO-BB3 sensor (Sea-Bird Scientific [WET Labs, Inc.]; serial number 349) set in a custom enclosure (Dall'Olmo et al. 2009).For each experiment, the AC-S and ECO-BB3 were connected by silicone tubing in a recirculating loop that was powered by a peristaltic pump ($ 60 mL min À1 ), similar to Poulin et al. (2018).The total residence volume of the recirculating loop was 3.8 liters and was initially filled at the beginning of each experiment with 0.2-μm filtered artificial seawater.At each sampling point, volume from each culture replicate was added to the recirculating loop in series using a 60-mL syringe.The volume of culture added was previously determined such that it would provide a backscattering signal at least 20% higher than background.A 0.2-μm Polycap capsule filter (Whatman), flushed with 50 liters of nanopure water prior to first use, was used to remove cells between sampling points.Optical data for each sample were logged over 1 min using the opensource software, Inlinino (Haëntjens and Boss 2020), and median values were used in subsequent analyses.

Derivations from optical measurements
Particulate absorption (a p ) and attenuation (c p ) were estimated using a calibration-independent technique in which measurements of 0.2-μm filtered sample water were subtracted from unfiltered measurements (Slade et al. 2010).A semiempirical scattering correction was applied to particulate absorption and attenuation spectra following Röttgers et al. (2013), which was shown by Kostakis et al. (2021) to consistently perform well over a wide range of water types and with less error propagation relative to other correction methods.Particulate scattering (b p ) was estimated as the difference between attenuation and absorption.Temporal variability in a p and c p are presented for the wavelengths measured by the ECO-BB3 (470, 532, and 660 nm), which also correspond to previous reports (Stramski and Reynolds 1993;Stramski et al. 1995;Durand and Olson 1996;DuRand and Olson 1998;Claustre et al. 2002;DuRand et al. 2002;Poulin et al. 2018).However, hyperspectral a p and c p data are also presented in Supporting Information Fig. S1.Chl a concentrations (mg m À3 ) were estimated from the particulate absorption line height at 676 nm (Chl a(676)LH ) following Roesler and Barnard (2013) using a chlorophyll-specific absorption value at 676 nm of 0.014 m 2 g Chl À1 (Boss et al. 2007 Chl a(676)LH was used to calculate C : Chl a ratios (weight : weight).Pigment analyses reported in the Supporting Information were conducted by unsmoothing and then decomposing particulate absorption spectra into component Gaussian functions following Chase et al. (2013).
Prior to each set of phytoplankton experiments, the ECO-BB3 was calibrated using the transmission-tube (c tube) of the AC-S and serial dilutions of 0.2-and 0.7-μm NIST-traceable polystyrene Nanosphere Size Standard beads (Thermo Scientific) in nanopure water.The volume scattering function (β) at 124 of the beads at different concentrations was measured and then theoretically modeled based on their particle size distribution and index of refraction using Mie theory (Sullivan et al. 2013).The calibration slope (m sr À1 count À1 ) of the ECO-BB3 was computed as the ratio between the theoretical slope (sr À1 ) and measured slope (counts m À1 ) of β and c.The particulate portion of the volume scattering function (β p , m À1 sr À1 ) was calculated by subtracting the contribution from 0.2-μm filtered seawater (Slade et al. 2010).The particulate backscattering coefficient (b bp ) for each wavelength measured by the ECO-BB3 was computed using the conversion factor χ = 1.12 derived from Sullivan and Twardowski (2009) as: Backscattering efficiency (b bp /b p ) was computed as the ratio of b bp to b p .Chlorophyll-specific, carbon-specific, and cell-specific optical coefficients were also calculated for particulate absorption, attenuation, and backscattering at 470, 532, and 660 nm.POC values used to estimate carbon-specific coefficients were derived from culture sample measurements, after accounting for dilution in the optical instruments.The Gordon parameter (g), which relates remote-sensing reflectance (R rs ) to inherent optical properties (Gordon et al. 1975), was approximated as: where a sw and b bsw correspond to pure seawater absorption (Röttgers et al. 2010) and backscattering (Zhang et al. 2009).

Statistical analyses
Nonparametric Spearman correlation coefficients were used to assess relationships between phytoplankton physiological measurements and phytoplankton bio-optical properties at 470, 550, and 660 nm.Correlation coefficients between 0 and 0.39 are described as weak, between 0.4 and 0.59 as moderate, and between 0.6 and 1 as strong.Only significant correlations (p-value ≤ 0.05) are discussed.
It is worth reiterating that in both T. pseudonana experiments, cultures were diluted to dawn cell concentrations at each sampling time point after the onset of cell division through dawn when cell division ended.Cell-specific FSC and SSC and red fluorescence increased during daylight and decreased in the dark (Fig. 2).
Minimum values of cell-specific scattering and fluorescence occurred at dawn, but the timing of maximum values varied depending on the property.FSC per cell showed a symmetrical bimodal pattern with maxima at peak irradiance and 2 h after dusk.SSC per cell exhibited a single maximum 2 h before dusk.Red fluorescence per cell displayed an asymmetrical bimodal pattern with maxima at peak irradiance and at the dusk transition.F v /F m displayed a symmetrical diel cycle, decreasing from dawn to dusk and then increasing from dusk to dawn (Fig. 3).
A diel pattern was apparent for POC and PON per cell, which increased from dawn to maxima at peak irradiance and again at dusk and then decreased during the night until 4 h before dawn (Fig. 4).
C : N followed similar patterns, but with maxima occurring 2 h before dusk.Chl a per cell increased until peak irradiance and then remained relatively constant before decreasing for the first 6 h of night and then increasing until dawn.C : Chl a increased in the daylight and decreased in the night.
a p displayed a symmetrical diel pattern, increasing in the daylight until 4 h before dusk and then decreasing through the night until dawn (Fig. 5).
There was some spectral dependence of a p , with higher values at 470 nm relative to both 532 and 660 nm.Strong correlations were observed between a p , c p , and concentrations of POC, Chl a, and cell abundance.a p was also strongly correlated with concentration-independent properties, including b bp /b p and proxies for cell size (FSC and SSC per cell and POC per cell), molecular composition (C : N and C : Chl), and pigment activity (red fluorescence per cell and F v /F m ) (Fig. 6).
Chlorophyll-specific a p increased for 6 h after dawn, then decreased until dusk and remained constant in the night.Cellspecific a p showed similar periodicity as a p , increasing for 6 h after dawn and then decreasing until 4 h before the next dawn (Fig. 7).
The diel periodicity in chlorophyll-specific a p was particularly pronounced at 470 nm, reflecting the strong absorption properties of Chl a and carotenoids at blue wavelengths (Morel and Bricaud 1986).Carbon-specific a p displayed an opposite trend as chlorophyll-specific and cell-specific a p , decreasing from dawn until peak irradiance and then remaining relatively constant before increasing 4 h after dusk (Fig. 7).Spectral variability in the range of each biomass-specific a p coefficient was greater than the temporal variability (Fig. 8).c p showed a similar diel pattern as a p , but with more asymmetry and little wavelength dependence (Fig. 5).c p exhibited a weaker correlation with chlorophyll concentrations than a p , but a stronger correlation with POC concentrations.c p was similarly correlated to phytoplankton physiological properties as a p , but showed stronger correlations with concentrationindependent properties, including proxies for cell size (FSC per cell, SSC per cell, and POC per cell) and molecular composition (C : N and C : Chl a).c p was also moderately to strongly correlated with pigment activity (red fluorescence per cell and F v /F m ) and b bp /b p (Fig. 6).Chlorophyll-specific and cell-specific c p increased in the daylight and then decreased at night, a pattern opposite that of carbon-specific c p (Fig. 7).Temporal variability in the range of each biomass-specific c p coefficient was greater than the spectral variability (Fig. 8).
b bp showed a bimodal pattern over 24 h, with an overall decreasing trend that precluded any significant relationships between b bp and measured metrics of phytoplankton physiology (Figs. 5, 6).b bp did increase from 13 : 00 and 17 : 00 on the 1 st day, consistent with previously described daytime increases in b bp (Poulin et al. 2018).Indeed, b bp and c p were strongly correlated from 13 : 00 on the 1 st day to 09 : 00 on the following day (r ≥ 0.81, p < 0.01 for 470, 532, and 660 nm), supporting the link between phytoplankton and diel b bp cycles.The basis of the discrepancy between b bp and c p between 09 : 00 and 11 : 00 on the 1 st day remains unclear.Chlorophyll-specific b bp and cell-specific b bp were also spectrally independent and decreased over the course of the experiment (Fig. 7).Comparatively, carbon-specific b bp followed a pattern akin to that of carbon-specific a p and c p , albeit with greater symmetrical periodicity.Like c p , each biomassspecific b bp coefficient showed greater temporal variability than spectral variability (Fig. 8).
Backscattering efficiency (b bp /b p ), a concentration-independent proxy for cell size and intracellular composition, displayed  an asymmetric diel pattern without spectral dependence (Fig. 5).b bp /b p decreased between dawn and 4 h before dusk and it then remained relatively constant for the following 8 h before increasing during the night.Moderate to strong correlations were observed between b bp /b p and a p , c p , as well as concentrationindependent proxies for cell size (FSC and SSC per cell and POC per cell), molecular composition (C : N and C : Chl a), and pigment activity (red fluorescence per cell and F v /F m ) (Fig. 6).

Cyanobacterium
Synechococcus doubling time was $ 3.2 d, with cell division beginning near peak irradiance and ending at the light-dark transition, as previously reported (Waterbury 1986;Armbrust et al. 1989) (Fig. 1; Table 1).In both experiments, FSC per cell increased during the day and decreased in the night, with a maximum 2 h before dusk and minimum at dawn (Fig. 2).SSC per cell and yellow fluorescence per cell were relatively invariant over the day-night cycle.F v /F m displayed a diel pattern that varied inversely with irradiance and exhibited a maximum 4 to 6 h after dusk (Fig. 3).
POC and PON per cell were constrained between 3-5 pg C per cell and 0.5-1 pg N per cell, yet still showed a diel pattern of increasing until peak irradiance, decreasing to dusk, and then remaining relatively constant at night (Fig. 4).The C : N diel cycle increased from dawn to 2 h before dusk and then decreased at night.Chlorophyll per cell was higher for the 1 st experiment relative to the 2 nd , but in both experiments decreased from near peak irradiance until dusk and then  remained relatively constant throughout the night.C : Chl a exhibited a similar diel cycle as C : N. A diel pattern was absent for a p , which generally increased from the beginning to the end of each experiment (Fig. 5).Moderate to strong correlations were observed for a p with c p , b bp , POC, Chl a, and cell concentrations, as well as concentration-independent properties of POC per cell, SSC per cell, and chlorophyll per cell (Fig. 6).Chlorophyll-specific and cell-specific a p only varied over a small range (as in Stramski et al. 1995) and showed similar wavelength dependencies as a p .However, some diel periodicity was apparent at 470 and 532 nm corresponding to the absorption properties of Chl a and carotenoids in the blue wavelengths and phycoerythrin in the green wavelengths (Fig. 7).Between the two Synechococcus experiments, a maximum in cell-specific a p 2 h before peak irradiance was observed in the 1 st experiment that was absent in the 2 nd .Both experiments showed a decrease in cell-specific a p from peak irradiance to dusk followed by little change at night.Carbon-specific a p exhibited a diel pattern at 470, 532, and 660 nm, decreasing from dawn to dusk and then increasing in the night (Fig. 7).Median chlorophyllspecific a p showed wide spectral variability, being higher at 470 nm than for 532 and 660 nm.Cell-and carbon-specific a p also showed greater spectral than temporal variability, in which median values at 470 nm were up to fivefold higher than those at 532 and 660 nm (Fig. 8).
c p showed spectral dependence, being highest at 470 nm and lowest at 660 nm and was positively correlated to changes in cell abundance (Fig. 5).For the 1 st experiment, c p increased from dawn to a maximum at dusk.For the 2 nd experiment, c p increased from dawn to 2 h before dusk.In both experiments, c p remained relatively constant over the night.c p was strongly correlated with b bp and concentrations of chlorophyll, POC, and cell concentrations (Fig. 6).c p was also moderately to strongly correlated with concentration-independent properties, including b bp /b p , proxies for cell size (FSC and SSC per cell), and molecular composition (C : N, C : Chl a, and Chl a per cell).Chlorophyll-specific c p displayed symmetrical diel periodicity for both experiments, increasing from a minimum at dawn to a maximum 2 h before dusk, and then decreasing over the course of the night (Fig. 7).In contrast, cell-specific c p increased from dawn until a maximum at peak irradiance and then decreased until 2 h after dusk before settling on a relatively constant value for the remainder of the night.Carbon-specific c p displayed a muted diel cycle with constrained spectral dependence relative to other biomass-specific coefficients.Nevertheless, all biomass-specific c p coefficients displayed greater spectral than temporal variability, with highest values at 470 nm and lowest values at 660 nm (Fig. 8).b bp displayed spectral dependence opposite that of c p , with highest values at 660 nm and the lowest values at 470 nm (Fig. 5).b bp was also moderately to strongly correlated with cell abundance, POC concentration, and chlorophyll concentration as well as concentration-independent properties of SSC per cell, and chlorophyll per cell (Fig. 6).Diel variability was absent in chlorophyll-specific b bp , while cell-specific b bp displayed a similar pattern to cell-specific a p and c p .Carbonspecific b bp exhibited a temporal pattern similar to carbonspecific a p and opposite that of carbon-specific c p (Fig. 7).Biomass-specific b bp coefficients varied spectrally and temporally (Fig. 8).
b bp /b p exhibited a symmetrical diel pattern of decreasing between dawn and dusk and increasing from dusk to dawn, with absolute values varying with wavelength (Fig. 5).b bp /b p was correlated with a p at 532 nm, c p , cell abundance, and chlorophyll concentrations, as well as concentration-independent physiological metrics for cell size (FSC and SSC per cell), and molecular composition (C : N, C : Chl a, and Chl a per cell) (Fig. 6).

Picoeukaryote
O. lucimarinus had a doubling time of $ 1.9 d, dividing predominantly in the dark, as is often the case for eukaryotic phytoplankton (Nelson and Brand 1979).Cell division was minimal in the 1 st experiment and more apparent in the 2 nd experiment (Fig. 1, Table 1).In both experiments, scattering per cell and red fluorescence per cell displayed diel patterns of daytime increases and nighttime decreases.However, the timing of maximum values for these parameters were offset between the two experiments, peaking at dusk for the 1 st experiment and 2 h before dusk for the 2 nd experiment (Fig. 2).F v /F m displayed a narrow diel cycle for both experiments that varied inversely with irradiance and remained constant at night.
POC and PON per cell were low and constrained between 0.17-0.29 pg C per cell and 0.03-0.06pg N per cell, showing a diel pattern of daytime increase and nighttime decrease (Fig. 4).C : N exhibited a small range and a muted diel cycle with an increase at the light-dark transition.Chlorophyll per cell was similar between the two experiments, increasing in the day and decreasing at night.C : Chl a indicated an accumulation of carbon relative to Chl a from the beginning to the end of the experiment and showed a diel pattern of increasing from 2 h after dawn until 4 h after dusk.
a p was spectrally dependent, with highest values at 470 nm and lowest at 660 nm (Fig. 5).a p increased over the duration of both experiments, displaying the highest increases of up to a factor of 2 in the 2 nd experiment.Strong correlations were observed between a p , Chl a and cell concentrations.a p was also correlated with concentration-independent properties of C : Chl, SSC per cell, F v /F m , and b bp /b p (Fig. 6).Chlorophyllspecific a p showed little diel variability, while cell-specific a p displayed diel periodicity of increasing from dawn to dusk and decreasing from dusk to dawn.Carbon-specific a p exhibited a low diel variability, decreasing from 2 h after dawn until 4 h after dusk and then increasing into the following morning (Fig. 7).Biomass-specific a p was spectrally dependent, with highest values at 470 nm and lowest at 532 nm (Fig. 8).c p showed the same spectral dependence as a p and displayed a diel cycle, increasing from dawn to dusk and decreasing from dusk to dawn (Fig. 5).The range in c p was similar between both experiments.c p exhibited a weaker correlation with cell abundance and chlorophyll concentration than a p .c p was correlated with b bp as well as with concentrationindependent properties of b bp /b p , cell size (FSC and SSC per cell and POC per cell), and pigment activity (red fluorescence per cell and F v /F m ) (Fig. 6).Both chlorophyll-and cell-specific c p showed the same symmetrical diel periodicity as c p .Carbon-specific c p displayed a stepwise decrease 4 h after dusk (Fig. 7).Each biomass-specific c p coefficient displayed greater spectral than temporal variability (Fig. 8).
b bp displayed a spectral dependence that differed from that of a p and c p , with the highest values at 532 nm and the lowest at 660 nm (Fig. 5).b bp was not correlated with cell abundance but was correlated with POC concentrations and concentration-independent properties of C : Chl and red fluorescence per cell (Fig. 6).Chlorophyll-and cell-specific b bp did not show clear diel patterns.Carbon-specific b bp exhibited a similar diel pattern to carbon-specific c p (Fig. 7).Biomass-specific b bp coefficients displayed less spectral variability than a p or c p (Fig. 8).
b bp /b p exhibited a symmetrical diel pattern and its magnitude depended on wavelength.b bp /b p decreased between dawn and dusk and then increased from dusk to dawn for both experiments (Fig. 5).b bp /b p was correlated with a p , c p , cell abundance, chlorophyll concentrations, as well as concentrationindependent proxies for cell size (FSC and SSC per cell, POC per cell), chlorophyll per cell, and red fluorescence per cell (Fig. 6).

Discussion
Bio-optical properties are thought to vary to first order with particle concentration and thus could be expected to follow cell division patterns.Each of the studied phytoplankton exhibited synchronized cell division coupled to the light-dark cycle, but they differed in timing and duration (Fig. 1).Diel changes in a p were linked to changes in cell concentration for all three species, but associations between changes in c p and b bp with cell concentration were generally restricted to Synechococcus (Fig. 8).Changes in c p and b bp over the diel cycle may thus be more sensitive to physiological properties, including cell size and refractive index, that change more continuously than more stepwise increases in cell concentrations (Stramski and Reynolds 1993;Stramski et al. 1995;DuRand and Olson 1998;DuRand et al. 2002).The suite and strength of correlations between each bio-optical property and metrics of cell physiology varied spectrally and by species (Fig. 8).Despite interspecific differences in timing and magnitude, diel cycles in physiological properties and bio-optical properties are often likely to correspond to photoacclimation, daytime accumulation of energy and carbon, and nighttime metabolism and respiration (Figs. 4,8).

Changes in cell physiology attuned to the day-night cycle
The phytoplankton studied here all showed diel changes in chlorophyll per cell, pigment absorption ratios, F v /F m and fluorescence per cell that corresponded to changes in light intensity.Photophysiological responses to both excess and low light were evidenced by changes in pigmentation, particularly ratios of photoprotective carotenoids and accessory pigments to Chl a (Supporting Information Fig. S2), F v /F m minima at peak irradiance for Synechococcus and O. lucimarinus (Fig. 3.), and daytime increases and nighttime decreases in red fluorescence per cell for T. pseudonana and O. lucimarinus (Fig. 4).
Phytoplankton diel periodicity in cell size and intracellular composition were strongly associated with changes in cellular optical properties that alter the bulk properties of seawater.In all species, daytime increases in cell size due to carbon accumulation from photosynthesis were reversed by cell division and respiration (Fig. 4).These behaviors were captured by diel changes in cell-specific forward light scattering and c p in all species (consistent with Olson et al. 1989;Olson et al. 1990;DuRand et al. 2002) as well as similar changes in cell-specific SSC in the eukaryotic cells (Figs. 1, 2).This is consistent with previous observations for picoeukaryotes in which diel changes in cell-specific SSC corresponded to changes in cell size (Simon et al. 1994;Vaulot and Marie 1999).b bp /b p has been suggested as a sensitive proxy for the bulk refractive index of natural particle assemblages (Ulloa et al. 1994;Twardowski et al. 2001;Boss et al. 2004), and it has been observed to increase with decreasing cell size in different phytoplankton taxa (Vaillancourt et al. 2004;Whitmire et al. 2010).For all three phytoplankton studied here, diel changes in b bp /b p reflected cell growth and division and/or respiration (Figs. 4,5).In addition, C : N ratios for each phytoplankton showed the expected diel pattern in which carbon-rich compounds (e.g., carbohydrates) accumulate faster than nitrogen-rich compounds (e.g., proteins) during the day, which is then reversed at night due to protein synthesis and respiration (Halsey and Jones 2015) (Fig. 4).
Sensitivity of absorption, attenuation, and backscattering to diel changes in physiology a p is primarily impacted by cell concentration, composition, and concentration of intracellular pigments, the efficiency of absorption by those pigments, and cell size (Morel and Bricaud 1981;Bricaud et al. 1983).Ahn et al. (1992) noted that variations in chlorophyll-specific a p are the product of interspecific differences in cell size and intracellular pigment concentrations.Thus, both interspecific differences and diel patterns in biomass-specific a p observed in this study likely reflect changes in pigmentation (concentration and composition) and cell size (Fig. 7).Larger cells and cells with higher internal concentrations of pigments have been hypothesized to exhibit a strong "package effect," that is, decreased absorption per unit of pigment (Morel and Bricaud 1981).This could be supported by this study with T. pseudonana exhibiting lower chlorophyll-specific a p than Synechococcus and O. lucimarinus.Furthermore, the relatively invariant chlorophyll-specific a p at 660 nm for Synechococcus and 470, 532, and 660 nm for O. lucimarinus may reflect a minimal contribution of a "package effect" at these wavelengths for these species.As T. pseudonana and Synechococcus cells increase in size (e.g., FSC per cell or POC per cell), chlorophyll-specific a p increases.The corresponding increases in cell size and chlorophyll-specific a p for these species is contrary to the prediction of a "package effect" and may be attributed to coincident increases in intracellular carotenoid concentrations, highlighting the complexity of the relationship between phytoplankton physiology and light.Furthermore, cellspecific a p for T. pseudonana, Synechococcus, and O. lucimarinus increased during periods of cell expansion.(Fig. 7).T. pseudonana exhibited diel variability in the Gordon parameter (g) at 470 nm, with decreases in the day due to increasing cell size and pigment absorption and increases at night due to division (Supporting Information Fig. S3).Comparatively, g did not show diel variability at 532 and 660 nm for T. pseudonana nor at 470, 532, and 660 nm for either Synechococcus or O. lucimarinus.These results suggest that changes in reflectance in regions of chlorophyll absorption are more impacted by diel cycles of species like T. pseudonana than species like Synechococcus and O. lucimarinus.c p is the sum of a p and b p and is thus impacted by pigmentation, cell size, and refractive index (Kirk 1975;Jerlov 1976).c p was dominated by scattering processes and consequently, was likely more sensitive to changes in cell size and refractive index than to changes in pigmentation as the average contribution of a p to c p was 5-19% for the phytoplankton species studied here.c p can increase with cell concentration if cell size remains constant or with cell size if cell concentration remains constant (Kirk 1975).T. pseudonana followed these expectations, showing inverse patterns between cell-and carbon-specific c p , with cell-specific c p increasing with cell size in the day prior to division and carbonspecific c p increasing following division at night (Figs. 4,7).Diel changes up to 2.75-fold in the carbon-specific c p coefficient contrast with previous observations of relatively constant carbon-specific c p over the day-night cycle for T. pseudonana (Stramski and Reynolds 1993).Synechococcus and O. lucimarinus also showed increases and decreases in cell-specific c p corresponding to increases and decreases in cell size.Synechococcus, however, showed little to no change in cell-specific c p or carbon-specific c p during the night when division was arrested and changes in POC, POC per cell, and Chl a per cell were minimal (Figs. 4 and 7).While c p and b bp can be well correlated across their full dynamic ranges in the ocean (Dall'Olmo et al. 2009;Westberry et al. 2010), they can also be decoupled in timing and amplitude over the diel cycle (Loisel et al. 2011;Kheireddine and Antoine 2014;Poulin et al. 2018).In the current study, c p and b bp followed strongly correlated temporal patterns in Synechococcus and O. lucimarinus, but c p showed amplitude changes $ twofold higher than b bp across all species (Fig. 5).For Synechococcus and O. lucimarinus, c p correlated with each physiological metric covarying with b bp , but c p also showed correlations with other physiological metrics not covarying with b bp (Fig. 6).These relationships may help to explain the temporal variability of biomass-specific c p and b bp .For example, opposite patterns between carbon-specific b bp and c p were exhibited in Synechococcus.Changes in carbonspecific c p for Synechococcus intuitively follow changes in cell size.By contrast, the decreasing carbon-specific b bp during the day may reflect the lower backscattering efficiency (b bp /b b ) of expanding cells while the increasing carbonspecific b bp in the night may reflect the higher b bp /b b of cells that lose volume from division and respiration (Figs. 5,7).

Comparison of normalized bio-optical coefficients with previous studies
Field and modeling studies often assume constant carbonspecific attenuation coefficients (Siegel et al. 1989;Stramska and Dickey 1992) or chlorophyll-specific absorption coefficients (Morel 1991) over the day-night cycle when converting optical measurements into POC concentrations and primary production rates.In this study, interspecific variability in normalized bio-optical coefficients decreased when the dynamic range of the properties they were normalized to also decreased.Specifically, variability was highest when biooptical coefficients were normalized to cell concentration, less when normalized to Chl a, and the least when normalized to carbon (consistent with observations by Stramski and Reynolds 1993;Vaillancourt et al. 2004) (Fig. 8).Thus, POC concentrations and primary production rates can be derived from in situ and remote optical measurements using constant carbon-specific bio-optical coefficients, but caution should be exercised before making the same derivations using constant cell-or chlorophyll-specific bio-optical coefficients.
Interspecific variability for each bio-optical coefficient was greater than diel variability for a single species, in agreement with previous studies (DuRand et al. 2002;Poulin et al. 2018).Across all three species over the diel cycle, the median chlorophyll-specific c p coefficient at 660 nm (169 AE 48 m 2 g Chl a À1 ) fell within the range reported over various areas of the open ocean (40-1420 m 2 g Chl a À1 ) (Behrenfeld and Boss 2003), while the median chlorophyll-specific b bp coefficient at 532 nm (1.2 AE 0.5 m 2 g Chl a À1 ) fell at the low end of the range reported for 9 cultured phytoplankton species by Ahn et al. (1992) (0.023-9.9 m 2 g Chl a À1 at 550 nm) and 13 different species by Whitmire et al. (2010) (0.3-463 m 2 g Chl a À1 at 550 nm).Cell-specific and carbon-specific b bp for all species showed a similar range to those reported by Poulin et al. (2018).At 660 nm, T. pseudonana showed the highest cell-specific c p coefficients, with values similar to previous studies (Stramski et al. 1995;Poulin et al. 2018).By contrast, cell-specific c p values for Synechococcus were $ 4 times lower than T. pseudonana, yet still $ 10 times higher values than values previously reported for another Synechococcus strain (WH8103) (Stramski et al. 1995).O. lucimarinus had cellspecific c p values a factor of $ 55 lower than T. pseudonana.Similarly, carbon-specific c p values were highest for T. pseudonana and lowest for O. lucimarinus, with the overall mean value at 660 nm (2.7 AE 1.6 m 2 g C À1 ) being comparable to those reported by previous diel studies including for T. pseudonana (3.81 m 2 g C À1 ) (Stramski and Reynolds 1993), Synechococcus (2.48 m 2 g C À1 ) (Stramski et al. 1995), Nannochloris (3-4 m 2 g C À1 ) (DuRand and Olson 1998), and M. pusilla (2.5-3.8 m 2 g C À1 ) (DuRand et al. 2002).

Conclusion
Variations in phytoplankton bio-optical properties are intertwined with diel changes in cell morphology, composition, and photophysiology.The current study provides the first culture-based characterization of complete diel cycles in b bp and, for the three species examined here, observed daynight changes ranged between 1.2-and 3-fold.Time of day is thus an important factor to consider when interpreting satellite-retrieved data.Diel changes in many bio-optical properties also differed in timing and magnitude between species, emphasizing the additional importance of phytoplankton community composition in interpreting remotely detected optical properties.

Fig. 1 .
Fig. 1.Photosynthetically active radiation (PAR) over the 12 : 12-h light-dark sinusoidal light cycle for each phytoplankton culture (a).Mean cell abundances and their standard deviations for each diel experiment (b) measured directly from the cultures (left) and after dilution in the recirculating optical sampling system (right).

Fig. 2 .
Fig. 2. Cell-specific scattering and fluorescence in the recirculating optical sampling system for each diel experiment.Thalassiosira pseudonana and Ostreococcus lucimarinus cells were enumerated using red fluorescence (chlorophyll indicator), while Synechococcus were enumerated using yellow fluorescence (phycoerythrin indicator).Red fluorescence data for Synechococcus are not shown as the flow cytometry settings for the red channel were set too low to resolve cell fluorescence from noise.Values represent population means normalized to the initial condition, with error bars representing the standard deviation.

Fig. 3 .
Fig. 3. Variable to maximum fluorescence (F v /F m ) of cells in cultures for each diel experiment.

Fig. 4 .
Fig. 4. Carbon, nitrogen, and Chl a per cell, as well as carbon to nitrogen (C : N) and carbon to Chl a (C : Chl a) ratios, estimated for the recirculating optical sampling system for each diel experiment.Values represent means with error bars as standard deviations.

Fig. 5 .
Fig. 5. Temporal changes over the day-night cycle of phytoplankton bio-optical coefficients at 470, 532, and 660 nm, including absorption (a p ), attenuation (c p ), backscattering (b bp ) as well as backscattering efficiency (b bp /b p ). Error bars represent standard deviations from mean values.

Fig. 7 .
Fig. 7. Biomass-specific bio-optical properties at 470, 532, and 660 nm over the diel cycle for each phytoplankton and each experiment.Bio-optical coefficients include those normalized to Chl a concentrations, to cell concentrations, and to particulate organic carbon concentrations.Error bars represent standard deviations from mean values.

Fig. 8 .
Fig. 8. Box and whisker plots comparing biomass-specific bio-optical properties at 470, 532, and 660 nm between each phytoplankton.Boxes represent the median and the 1 st and 3 rd quartiles.Whiskers represent the 1.5 interquartile range.Circles indicate outliers.

Table 1 .
Overview of cultures, replicates, and sampling scheme for each experiment.Mean growth rates (μ) and their standard deviations are reported for culture data following acclimation to growth conditions, including those taken for the diel experiments.Cell concentrations are reported for both the cultures and the recirculating sampling loop.At each sampling point of each experiment, volume from each culture replicate was diluted into the 3.8 liters residence volume of the recirculating sampling loop connecting the optical instruments.