Diatom cell-specific silica production as a morphological adjustment to high growth irradiance

Marine diatoms represent one of the most successful phytoplankton groups in the marine environment. Their ecological success has been attributed to their photosynthetic plasticity, and the protection afforded them by their rigid silica cell walls. In this study, we investigated the influence of light environments on the physiology and morphology of the centric diatom Thalassiosira oceanica. Specifically, we determined how irradiance influenced growth rates, cell stoichiometry, photobiology and silica production. We found that T. oceanica exhibited optimal growth at irradiance around 250 µmol photons m -2 s -1, and showed a 15%–23% increase in frustule thickness at growth irradiances at or above 480 µmol photons m -2 s -1. While we found no direct relationship between growth and silicification, we did find that heavier silicified cells showed reduced dependence on non-photochemical quenching and lower excitation pressure when exposed to high irradiances. The observed co-adjustment of cell properties and silicification in response to growth irradiance, suggests that T. oceanica uses both physiological and morphological plasticity in photo-acclimation. These findings shed light on the functionality of frustule morphology and plasticity on diatom photo-acclimation, whereby through coordinating cellular changes with frustule properties, T. oceanica may be able to modulate the intracellular light field of its cell to optimise growth. This light-induced frustule modulation may represent an important growth strategy for natural diatom assemblages found in environments where light exposure can be highly dynamic.

Diatoms have evolved to dominate phytoplankton communities across much of the ocean. Their rise to ecological success has been attributed to their photosynthetic flexibility and capacity for photoprotection in mixed and often turbulent waters (Lavaud, 2007; Depauw et al., 2012). Unique amongst the phytoplankton, diatoms require silicic acid to produce their opal cell walls, known as frustules, an essential element for their reproduction and binary division. This morphological trait is believed to have played a part in their success in contemporary oceans, where the dense mineralised frustule provides the cell with ballast (Raven and Waite, 2004), protection from grazers (Hamm et al., 2003; Pondaven et al., 2007; Petrucciani et al., 2022), and can modulate light transmission, either through light focussing (De Stefano et al., 2007) or light deflection (Yamanaka et al., 2008; Ellegaard et al., 2016). Today, due to their physiological plasticity and unique morphology, diatoms are key autotrophs in the ocean, responsible for a large proportion of global marine primary production and carbon export (Field et al., 1998), and as mostly obligate silicifiers, are central to ocean silicon cycling (Tréguer et al., 2018).

Light and dissolved nutrients are the primary determinants of growth and photosynthesis in diatoms, and the physiological plasticity of a given species will ultimately determine its ability to survive and thrive under varied or fluctuating conditions. In the global ocean, light is generally dynamic, meaning that diatoms need ways to maximise their energy capture for growth and photosynthesis, yet balance that through photoinactivation, and avoiding photoinhibition or photodamage (Campbell and Serôdio, 2020). In instances where light is highly variable, this is typically done via rapid adjustment of the photosynthetic machinery to optimise energy input with metabolism (Arsalane et al., 1994; Giossi et al., 2025). For high light environments or when mixing brings diatoms to the surface, photoprotective mechanisms are initiated. In diatoms, the first line of defence under high irradiance is typically non-photochemical quenching (NPQ), which involves the rapid dissipation of excess heat via the de-epoxidation of antenna pigments and light harvesting proteins (Arsalane et al., 1994; Ruban et al., 2004; Croteau et al., 2025). However, if these dissipating processes are insufficient, photoinhibition may ensue, and if not resolved, eventually damage to the photosynthetic machinery will occur (Lavaud et al., 2004).

Unlike rapid adjustment processes to changes in light, longer-term exposure to different irradiances results in cellular changes to acclimate the cell to the enduring light conditions. Photomechanistic adjustments to low light in diatoms include enlarged light-harvesting antenna size, and a larger thylakoid membrane area, which together house higher chlorophyll pigment concentrations, ensuring high light-harvesting capacity (Wilhelm et al., 2014). Although this strategy maximises photochemistry for low light conditions, exposure of these chlorophyll dense cells to high light will cause a rapid imbalance between absorbed energy and its metabolic use, causing oxidative stress or damage. In cells acclimated to high irradiances, antenna size, thylakoid membrane area and chlorophyll content are greatly reduced (Wilhelm et al., 2014), this helps cells avoid over-excitation and damage by reducing the risk of self-shading through densely packed pigments (“packaging effect”), and substantially increases the absorption efficiency per unit chlorophyll. While many structural and functional aspects of light acclimation have been minutely described in diatoms, one relatively understudied trait that may contribute to diatom photo-acclimation is their biomineralised silica frustule.

Diatom frustule morphology and silica incorporation have been shown to vary with nutrients (Durkin et al., 2012), pH (Petrou et al., 2019; FitzGerald-Lowry et al., 2025), temperature (Sheehan et al., 2020; Baker et al., 2016), zooplankton grazing (Petrucciani et al., 2022; Pondaven et al., 2007) and light (Su et al., 2018b; FitzGerald-Lowry et al., 2025), attesting to the phenotypic plasticity of this trait in response to environmental conditions. In the case of light, both the quality and quantity of light have been shown to alter cell size, frustule pore density and pore diameter (Su et al., 2018a), where higher irradiances increase valve thickness, while less dense or thinner frustules are produced under suboptimal light or at particular wavelengths (Su et al., 2015). Detailed studies on the optical properties of diatoms have shown how nanofeatures can modify light, highlighting the capacity for the frustule to either filter or focus light depending on foramen pore size (Maibohm et al., 2016; Ellegaard et al., 2016; Goessling et al., 2024). If frustule nanostructure can influence the quality and quantity of light entering the cell, and frustule formation is phenotypically plastic, it holds that the structural-optical properties of the diatom frustule have the potential to influence photosynthesis. Indeed, experimental data have shown how differences in diatom frustule nanostructure can modulate the intracellular light field of a cell via photon trapping and forward scattering processes, and that these are coupled to the chloroplastic properties within the cell (Goessling et al., 2018). However, the functionality of frustule morphology and plasticity upon diatom physiology remains underexplored. Here, we assess relationships between photosynthesis, photoprotection and silica incorporation rates as a function of growth irradiance to determine whether acclimated growth irradiance influences diatom cell-specific silica incorporation, and if the frustule plays a part in diatom photo-acclimation.

Culturing and maintenance

The temperate, centric (non-chain forming) diatom, Thalassiosira oceanica was isolated from surface seawater collected from the Port Hacking oceanographic time series station (34.1192 °S, 151.2267 °E) in September 2020. Prior to experimentation, cultures were maintained in F/20 (1:9 dilution of F/2) medium (Guillard and Ryther, 1962; Guillard, 1975) at 20 °C for 4 weeks and acclimated either to high light (200 µmol photons m -2 s -1) or low light (30 µmol photons m -2 s -1) conditions on a 12:12 light:dark cycle.

Experimental design

For the experiment, cultures of low and high light acclimated T. oceanica were grown in 1L Schott bottles filled to 500 mL with F/20 and inoculated with stock culture to a final concentration of 5,000 cell mL -1). Culture bottles (n = 24) were divided between six temperature-controlled tanks maintained at 20 °C. The irradiance levels for each tank were 25, 60, 120,250, 480, and 700+ µmol photons m -2 s -1 (Tanks 1–6) supplied on a 12:12 light:dark cycle. For the three lowest light levels (Tanks 1–3), inoculum of the low light-adapted T. oceanica was used, whereas the higher-light adapted T. oceanica was used for the three highest light levels (Tanks 4–6). For the highest light treatment (Tank 6), replicates (n = 4) were first acclimated in 480 µmol photons m -2 s -1 (Tank 5) for 1 week before being moved into the highest light treatment to help with adjustment. Subculturing was conducted weekly (on actively growing cultures) over a minimum of three consecutive weeks before measurements commenced.

Cell abundance and growth rates

Following acclimation (3–4 weeks) to their respective light levels, T. oceanica cell density was measured daily over 9 days to determine growth rates. Daily sub-samples (1 mL) were fixed in glutaraldehyde (1% final concentration) and stored at 4 °C for later counting. Counts were obtained via flow cytometry (Cytoflex, Beckman and Coulter) as events per microlitre. All counts were performed in technical triplicate. Growth rates were determined using a ‘Baranyi’ model fit generated using the R package MicrobialGrowth (L'Homme and Breard, 2025). All physiological parameters were measured during the following growth cycle, once cultures had reached exponential phase.

Chlorophyll a fluorescence

The photophysiological condition of T. oceanica acclimated to a range of growth irradiances was assessed using a Pulse Amplitude Modulated (Water PAM, Walz GmbH, Effeltrich, Germany) fluorometer. Prior to measurements, T. oceanica cultures were dark adapted (∼10 min) in low ambient light (<1 µmol photons m -2 s -1) before being transferred to a quartz cuvette, where an eight-step light curve of 1 min duration for each actinic light level (0, 163, 240, 360, 540, 764, 1,050, 1,690, 2,450 μmol photons m -2 s -1) was performed. Minimum and maximum fluorescence were recorded for each light step to obtain the following parameters: maximum quantum yield of PSII (F v /F m), relative electron transport rate (rETR) and non-photochemical quenching (NPQ). Light curves were fitted using the fit_photosynthesis function from the R package “Photosynthesis” (Stinziano et al., 2020; Stinziano et al., 2021), based on a non-rectangular hyperbolic model (Marshall and Biscoe, 1980). If a drop in rETR was observed after a peak rate, photoinhibition was included in the model, otherwise the default model was used. Photosynthetic parameters, maximum relative electron transport rate (rETR MAX), light utilisation efficiency (α), minimum (E k) and maximum saturating irradiance (E m), were extracted from the model fit using linear regression. Non-photochemical quenching and excitation pressure (1-qP) parameters were calculated from the PAM output. Trade-offs in allocations of excitation energy to photochemical versus non-photochemical quenching pathways were calculated as the fraction of open RCIIs responsible for photochemical quenching [1-C] vs. dynamic non-photochemical quenching [1-Q] under each actinic light intensity following equations from Suggett et al. (2015) where the product of [1-C] and [1-Q] is equivalent to the photochemical efficiency of PSII normalised to F V /F M.

Chlorophyll pigment concentrations

For determination of chlorophyll concentrations, a 100 mL sub-sample from each replicate was filtered onto a GF/F filter (Whatman) before being placed in cryovials and flash frozen in liquid nitrogen and stored at −80 °C until extraction. Frozen filter papers were placed into falcon tubes and chlorophyll was extracted by adding 3 mL of 90% acetone, vortexing, and storing the tube in a dark refrigerator overnight. The tubes were centrifuged (4 min at 4,000 rpm), and the absorbance of the supernatant was measured using a spectrophotometer (Shimadzu UV-1280) at wavelengths of 630 nm and 664 nm. The concentrations of chlorophyll a and c 1 + c 2 were calculated in μg mL -1 using the following equations (Jeffrey and Humphrey, 1975; Ritchie, 2006).

Particulate organic carbon (POC) and nitrogen (PON)

To determine particulate organic carbon (POC) and particulate organic nitrogen (PON) content, subsamples (100 mL) were filtered onto pre-combusted (450 °C for 5 h) 47 mm Whatman GF/F filters (0.7 µm nominal pore size), and individual filters wrapped in pre-combusted aluminium foil. Samples were immediately stored at −30 °C until further processing. To prepare samples for analysis, filters were thawed, then acidified with 1M HCl (dropwise) until saturated and placed in a drying oven at 60 °C for at least 48 h. Dried filters were cut in half, wrapped and pelletised in 9 mm tin capsules (Elemental Microanlysis, UK), and stored in moisture-free container. Background concentrations were quantified and corrected for using triplicate ‘blanks’ (filters with 100 mL of F/20 medium) collected and prepared as above. Analyses were conducted by the Biogeochemical Stable Isotope Facility at the University of Hawai’i at Mānoa using a Costech Elemental Combustion System with a Zero Blank Autosampler (ECS 4010) coupled with a Thermo Scientific Delta V Advantage, connected through a Thermo Scientific Conflo IV isotope ratio mass spectrometer.

Cell specific silica incorporation rates using PDMPO

To determine silicification rates of T. oceanica, the fluorescent dye PDMPO was used. This dye is taken up and bound in the diatom frustule during silicification, rendering PDMPO fluorescence proportional to silicification over time (Shimizu et al., 2001). On the day of exponential growth for each specific treatment, small sample bottles were filled with 100 mL of F/20 media and aliquot of T. oceanica culture were added (final concentration 5,000 cells mL -1). Then PDMPO was added (final concentration 125 nM), and the time noted on the bottle before being returned to respective treatment tank for 24 h (Leblanc and Hutchins, 2005). Precisely 24 h later, subsamples (1 mL) from each PDMPO replicate were preserved with glutaraldehyde (final 1%) flash frozen in liquid nitrogen and stored at −80 °C.

Silica incorporation determined by fluorescence microscopy

Slides for fluorescence microscopy imaging were prepared as follows: Cells preserved in glutaraldehyde were pelleted using centrifugation (4000 RCF for 5 min), supernatant replaced with 10% HCl and cells resuspended to facilitate cell lysis and removal of unbound PDMPO from within the cells. Cells were subsequently re-pelleted and washed with miliQ to remove traces of HCl and to reduce salt content, before final pelleting and resuspension in 50 µL of miliQ water. Each cell suspension was transferred and dried onto poly-L-lysine coated microscopy slides, and any remaining salt precipitate washed off with a gentle stream of miliQ water. Once dry, ∼15 µL of ProLong™ Gold Antifade Mountant (ThermoFisher Scientific, Inc.) was added to each slide and samples sealed with a coverslip. Slides were left to cure for 24 h before imaging. It should be noted that this mounting medium is not recommended for diatom frustules due to a similar refractive index (∼1.5) making the diatoms difficult to locate visually, however, it is optimal for fluorescence imaging.

Fluorescence imaging was carried out an upright microscope (Nikon Eclipse Ni, Japan) using an mCherry filter cube for chlorophyll autofluorescence detection and a custom filter cube (AT375/28x, AT415dc, AT535/40m) for PDMPO. Cells were imaged at a total of ×400 magnification, and all samples were processed within 2 days. Microscopy images were cut out into smaller individual images, each containing 1 cell, and a custom script in ImageJ (Fiji) was used to automatically measure and remove background fluorescence (mean background fluorescence +1SD of n = 4 measurements) from each image, before measuring the fluorescence of the image as representing the fluorescence from the individual cell. If more than 1 cell was present in the image, the total fluorescence was divided by the number of cells in the image.

Image processing for cell size and silica incorporation

Width and length measurements made on imaged cells were used to calculate the area of the cell using the assumption that each cell expressed cylindrical morphology. Only cells seen from the girdle view were used for this calculation (rectangular view) and included for further analysis. Fluorescence (PDMPO) from individual cells were adjusted for apparent bleed over from unidentified components in the cell (non-silica related) by first calculating the area adjusted fluorescence (raw fluorescence/cell area) and then creating a linear model of the relationship between PDMPO fluorescence and the mCherry fluorescence channel and extracting the residuals. Using the residuals instead of the raw PDMPO values, we eliminated any relationship with the auto fluorescence channel and thus ensured any effects of treatment were not affected by non-silica/PDMPO related fluorescence.

Data analysis and statistics

All data analyses and statistics were carried out in R (v.4.5.3 (Team, 2026)). All data were tested a-priori for homogeneity of variance using Levene’s test from the package “car” (Fox and Weisberg, 2018) and normal distribution via visual inspection of histograms and Shapiro-Wilk normality test. Data that did not display homogeneity of variance across groups were transformed using log10 or square root depending on distribution. Linear Mixed-Effects models were used to test the effect of treatment between groups using the lmerTest package (Kuznetsova et al., 2017), with replicate as the random intercept. If the random intercept did not explain a significant proportion of the residual variance, model complexity was reduced to a simple linear model. Estimated marginal means was used to compute pairwise comparisons between each group based on the respective model, using the emmeans package (Lenth and Piaskowski, 2026), with “tukey” p-value adjustment for multiple comparisons.

Growth, cell size, pigments and elemental stoichiometry

Growth rates corresponded with growth irradiance, with the three highest light treatments showing the highest specific growth rates of around 1.7 days -1 (Figures 1A,B). Cultures growing at the lowest two irradiances grew significantly slower than all other treatments, while the cultures acclimated to 120 were significantly higher than the low light, but lower than the high light acclimated cultures (F 5,16 = 60.71, P < 0.001). Cells grown at the lowest irradiance had extremely low total yields, indicative of severe light limitation (Figure 1A), despite their specific growth rate determined as ∼0.5 days -1 (Figure 1B). For this treatment, rates were determined manually, as the sigmoidal model used for the other growth curves overfit the data. These rates were calculated based on the initial doubling of cells following the long lag phase, but the authors believe the rate to be an over-estimation, and therefore should be treated with caution. Cell sizes ranged from 2.3 to 9 µm (length) and 4.4–10.5 µm (diameter), however, we found no effect of growth irradiance on cell volume, although there was some indication of smaller cells in the cultures grown at 250 µmol photons m -2 s -1 (Figure 1C). As expected (Behrenfeld et al., 2008), both chl- a and chl- c followed a light-dependent response, with a significant decline in chl- a cell quotas (F 4,15 = 20.59, P < 0.001) with increased irradiance (Figure 1D). Chl- c was significantly higher in cells acclimated to 60 µmol photons m -2 s -1 (t = −3.639, P < 0.01) compared with the other treatments (Figure 1E). We found no difference in the chl- c: a across the light treatments (Figure 1F). The carbon to nitrogen (C:N) ratio showed a parabolic response across the light levels, with significantly higher C:N for cultures grown between 120 and 400 µmol photons m -2 s -1, with a peak close to Redfield (5.6:1) in cells grown at 250 µmol photons m -2 s -1 (F 4,15 = 16.25, P < 0.001; Figure 1G). Data for the lowest light treatment were excluded from the cell-density dependent measurements due to total sampled biomass being too low for reliable quantitation.

Photochemistry and photoprotection

We saw only moderate differences for the relative electron transport rate (rETR) vs. PAR between the light-acclimated cultures (Figure 2A). However, we found light-dependent differences in all other photophysiological parameters. Non-photochemical quenching (NPQ) showed strong differences across treatments with highest levels recorded for the two lowest light levels, clearly separating from the higher light treatments (Figure 2B). Maximum quantum yield of PSII (F V /F M), which was >0.70 at the two lowest light levels, declined to <0.55 in the three highest light levels (F 5,17 = 85.3, P < 0.001; Figure 2C). We found differences in light utilisation efficiency (α; F 5,17 = 24.63, P < 0.001; Figure 2D) and maximum rETR (rETR MAX; t = −5.650, P < 0.0001), with a minimal value for both measured in the cultures grown at an irradiance of 250 µmol photons m -2 s -1 (Figure 2E). Saturating irradiance (minimum-E K and maximum-E M) showed a light-dependent increase, with values for E K ranging from 550 to 820 µmol photons m -2 s -1 (t = 3.061, P = 0.008; Figure 2F) and E M ranging from 1,690 to 1850 µmol photons m -2 s -1 (F 5,10 = 9.547, P < 0.001; Figure 2G).

Looking deeper into the photochemistry, we found that the maximum energy directed towards photoprotection (NPQ MAX) ranged from moderate (0.25–0.31) in the low-light acclimated cultures, to negligible (<0.10) in high-light acclimated cultures (Figure 3A). There were also large differences in NPQ onset irradiance between cultures, where NPQ was activated immediately upon illumination in the two cultures acclimated to the lowest light levels, compared with NPQ activation at much higher irradiances in the moderate and high-light cultures (Figure 3B). Comparing the proportion of energy directed towards photochemistry vs. photoprotection with the stepwise increase in irradiance during the light curve, we saw that the dynamic non-photochemical quenching [1-Q] was greater in the cells grown at lower irradiances, yielding a steady decrease in [1-Q] with the progressive closure of RCII, seen as a decline in [1-C] (Figure 3C). The high-light acclimated cultures showed comparatively minimal change in response to progressive decrease in [1-Q]. Similarly, if we look at qP MAX vs. NPQ MAX (qP and NPQ determined at the highest irradiance level ∼2,450 µmol photons m -2 s -1) we find that the relationship between the two parameters is sigmoidal, where T. oceanica undergoes a steep phase of NPQ development within a narrow range of qP, suggestive of constitutive shifts in xanthophyll pigment content as growth irradiance shifts from 60 to 120 µmol photons m -2 s -1 (Figure 3C: inset). This steep phase is reflected also in the maximum excitation pressure (1-qP MAX) as a function of NPQ MAX in cultures across growth irradiances, where cells grown at lower irradiances were under much higher excitation pressure, despite higher, more dynamic NPQ than those acclimated to higher irradiances (Figure 3D). When viewed as a ratio of photochemistry to photoprotection at maximum irradiance (qP MAX:NPQ MAX), it becomes clear that the proportion of light energy being used for photochemistry is much greater in the high-light acclimated cultures than the cultures grown at low-light, but with no further increase beyond a growth irradiance of 250 µmol photons m -2 s -1 (Figure 3E).

Cell specific silicification

The area-adjusted cell specific silicification followed a light-dependent response, with higher per cell silica incorporation at higher irradiance (F 5,17 = 6.995, P = 0.001; Figure 4A). When correlated with growth rates however, we found that silica incorporation did not directly follow growth, instead there was an apparent growth/light threshold at which silica incorporation was upregulated (Figure 4B). The cell-specific silicification rate remained stable between 60 and 250 µmol photons m -2 s -1 before increasing 15%–23% at irradiances at some point above 250 µmol photons m -2 s -1. To determine if any relationships exist between silicification and photophysiological state, we tested whether the changes in excitation pressure (1-qP) and the maximum photochemistry to photoprotection ratio (qP MAX:NPQ MAX) showed any correspondence with silicification. We found weak yet significant relationships of both parameters with silicification across the light-acclimated cultures (P = 0.021, P = 0.029, respectively), with the regression models explaining approx. 20% of the variation (Figures 4C,D). The high-light acclimated cells which were more heavily silicified showed lower excitation pressure and higher proportion of photochemistry to photoprotection than the cells grown at low and moderate light levels.

Here we assessed the effect of irradiance on the growth rate, chlorophyll, photoprotective capacity and silica incorporation rates for the centric diatom Thalassiosira oceanica. We found that irradiance had a strong effect on specific growth rate, where the higher the irradiance the faster the cells grew reaching a threshold of maximum growth at or above 250 µmol photons m -2 s -1. This acceleration in growth was accompanied by declines in the chlorophyll content, but no notable or directional change in cell volume. This has implications for photochemistry, as cell size can influence the absorption efficiency of chlorophyll molecules (Wilhelm et al., 2014). In the absence of cell size differences, the higher chl- a content in the lower light grown cells, means that there was a greater pigment packaging effect and may account for the observed higher excitation pressure in the cells grown at the lowest irradiance when exposed to high light (Wilhelm et al., 2014). The ecological success of diatoms has been largely attributed to their ability to utilise efficient photosynthetic adjustment strategies that allow them to acclimate rapidly to changing light conditions (Wagner et al., 2006; Ruban et al., 2004), yet the (Ruban et al., 2004)NPQ measured in T. oceanica in this study was relatively low, a characteristic that has been previously attributed to genetic encoding based on the relatively stable light climate of its open-ocean habitat (Lavaud et al., 2007). Comparisons across diatom taxa revealed that T. oceanica lagged in initiating protective mechanisms and expressed low levels of NPQ compared with the other diatoms (Lavaud et al., 2007; Fisher et al., 2020). This low NPQ suggests that instead of rapid induction of NPQ, T. oceanica places greater reliance on the downstream processes of the light-dependent reactions to dissipate excess energy following PSII excitation. Despite this low NPQ, we found a strong influence of growth irradiance on photoprotective responses, indicative of light-induced structural changes within the cell. We detected a more dynamic photoprotective response in low-light acclimated cells, with rapid onset and higher maximal amplitude of NPQ, in line with general expectation (Lavaud et al., 2002). These cells also exhibited a higher NPQ at high excitation pressures, evidenced by the higher NPQ:1-qP ratio. The higher conversion of photons to excitation pressure in the low light acclimated cells, could reflect a larger PSII antenna size (Derks and Bruce, 2018), a conclusion that may be attested by the higher chl- a content and F V /F M values measured. In contrast to the low light cells, we measured a delayed and diminished NPQ response in the higher-light acclimated cultures, which could suggest that T. oceanica may downregulate excitation pressure under steady state conditions. It has been shown that at high irradiances, NPQ in diatoms can operate in concert with PSII electron cycling, providing an enhanced protective mechanism against excess light (Arsalane et al., 1994; Lavaud et al., 2002; Lavaud et al., 2004). Therefore, the divergent patterns in photoprotective versus photochemical regulation observed across the low-to high-light acclimated cultures may be a result of changes to energy dissipation via the PSII electron cycle, where dependence on PSII electron cycling is greater in the high-light acclimated cultures, explaining the low NPQ despite the highly reduced state of PSII (Ruban et al., 2004).

In relating shifts in frustule morphology with photoacclimation, these data lend support to the idea that frustule synthesis may be induced in diatoms as a means of modulating their intracellular light field. We saw up to 23% more silica incorporation per cell surface area in cells acclimated to irradiances above 250 µmol photons m -2 s -1, indicating that cells grown at these higher light levels possessed thicker valves and/or reduction in valve pore size. Frustule change in response to high growth irradiance has been seen before in the large centric diatom Coscinodiscus, where pore size was smaller in cells acclimated to 300 µmol photons m -2 s -1 compared with those grown at lower light intensity (Su et al., 2018a). When interrogating relationships between changes in physiological traits and frustule synthesis, we found no direct relationship with growth rate, but rather a step-wise jump in silicification at the two highest growth irradiances, suggesting these two traits are decoupled. This disconnect seems reasonable if we consider the time frames for frustule synthesis versus diatom growth. Diatom cell division is typically in the order of 1–2 days, whereas frustule nanostructure formation occurs within minutes, and frustule thickening (Coradin and Lopez, 2003) and cell wall biosynthesis within hours (Hildebrand et al., 2006). This mismatch in cell growth and frustule formation adds to a growing body of evidence that suggests silicification is not a passive consequence of slower growth, but more likely a regulated and plastic physiological process responding to environmental cues. This uncoupling leads us naturally to question whether changes in frustule morphology in response to light may instead be linked to photosynthesis via internal light modulation and specifically, photoprotection.

A focus of this study was to investigate changes in cell structural properties that may contribute to photosynthetic regulation. To this end, we found significant relationships between frustule synthesis and photoprotective properties, where the high-light acclimated cells which had the thickest frustules showed the lowest photosynthetic stress response (lowest excitation pressure and highest proportional energy use in photochemistry). This structural (frustule) shift by T. oceanica as a means of photoprotective regulation is supported by previous work that has demonstrated how changes in frustule pore size can filter incoming irradiance, even selecting for wavelengths (Ellegaard et al., 2016; Maibohm et al., 2016). It has also been shown that frustule nanostructures can modulate the intracellular light field of the cell via the interaction between scattering with chlorophyll absorption properties (Goessling et al., 2018). Given the two lines of evidence above, it could be argued that by increasing frustule thickness, the high-light acclimated cells in this study could be restricting the incoming irradiance, while their reduced chlorophyll content, which reduces the risk of overexcitation, enables the cell to maximise photochemistry via the propagation and homogenisation of the light within the cell across the pigment molecules. Of particular interest were the nuanced results of the cells grown at 250 µmol photons m -2 s -1. These cells shared characteristics with both low and high-light acclimated cells. The high growth rates, low chl- c content and photoprotective properties parallelled those of the two highest light acclimated cultures, yet their chl- a content and silica production were consistent with the low-light acclimated cells. These opposing structural characteristics offer a unique physiological profile to explore the frustule-derived light modulation further. If we follow the same argumentation used above, we can assume that the incoming irradiance is not being as heavily filtered (lower silica content) by these cells and therefore the light entering the cell is transferred to the interior and scattered, interacting with the densely packed chlorophyll molecules, which could results in overexcitation, as seen with the lower light acclimated cells. However, these cells shared photoprotective traits with the highest light acclimated cells, which suggests that our T. oceanica has a light optimum (highest growth rate and carbon content) somewhere around 250 µmol photons m -2 s -1, where the cells grown at lower irradiances were light limited (lower growth). In contrast, cells growing at irradiances above 250 µmol photons m -2 s -1 were able to maintain high growth rates and low excitation pressure through the co-adjustment of their chlorophyll content and frustule thickness. Where working in concert, this trade-off serves to modulate the intracellular light field of the cell.

While the observed relationship between frustule thickening and photoprotection support the role of the frustule as an optical shield, and in combination with the co-adjustment of chlorophyll content, may work to regulate the internal light environment (Goessling et al., 2018; Goessling et al., 2021), this shield hypothesis implies that increasing frustule thickness is a more energetically favourable strategy to reduce light input, than the continuous regulation of NPQ, a supposition yet to be verified. However, other mechanisms, which may not be mutually exclusive, could be proposed to explain increased frustule synthesis under high light. One possibility could be that silicification acts as an alternative structural sink for energy dissipation under high growth irradiance, where the cell avoids metabolic damage by channelling excess reduced energy into frustule biosynthesis, akin to the theory of ‘carbon overflow’ during nutrient limitation (Fogg, 1983; Wear et al., 2015). Notwithstanding the underlying mechanism, light-induced modulation of the silica frustule appears to be an important growth strategy for T. oceanica under high light conditions, and in considering the broader ecological implications, make our findings relevant for photoacclimation in diatoms exposed to dynamic light conditions. For T. oceanica in shallow mixed coastal ocean environments, adjustment of frustule architecture could represent an effective strategy to avoid photoinhibition or minimise damage from UV radiation (Ellegaard et al., 2016; Goessling et al., 2024). Equally, changes to frustule density through increased silica deposition, would increase ballast (Arrieta et al., 2020) promoting cells to sink below the mixed layer and escape the high surface irradiance (Raven and Waite, 2004).

The findings from this study have advanced our understanding of diatom morphological and physiology co-adjustment in response to growth irradiance, highlighting a potential role of the frustule in photo-acclimation. However, as is often the case, this study has raised more questions than it has provided definitive answers, where the links between frustule architecture and photo-acclimation while tractable, are yet to be fully resolved. The idea of there being a potential light threshold at which diatoms induce their frustule to aid in light modulation, while enticing, requires further investigation. To adequately disentangle these cause-and-effect mechanisms within diatom metabolism, future studies will need to address the challenges associated with separating light modulated biochemical (pigments) responses from cell structural (frustule) dynamics. In this way, we may be able to empirically verify that diatoms use their frustules to actively modulate their internal light environment to minimise photoinhibition.