Clonal culture isolates of the cosmopolitan diatom Leptocylindrus danicus used in this study were originally collected and established via single-cell isolation by Dr. P. Ajani between December 2015 to February 2016 (Table 1). For the present laboratory latitudinal study, strains were selected from four thermally distinct latitudes along the south-eastern Australian coastline, spanning from the north coast to south coast of New South Wales: Coffs Harbour (30.2986° S, 153.1094° E), Forster (32.1796° S, 152.5118° E), Maroubra (33.9495° S, 151.2437° E) and Twofold Bay (37.0844° S, 149.9277° E) (Table 1; Fig. 1). Prior to acclimation for experiments (described below), the established L. danicus strains from Coffs Harbour, Maroubra and Twofold Bay were maintained in 33 ppt salinity fresh seawater (FSW, 0.2 µm filtered) enriched and autoclaved with f/2 medium (Guillard and Ryther, 1962). Batch cultures were maintained in 50 mL culture flasks at 18 °C and illuminated at ∼35 µmol photons m⁻² s⁻¹ on a 12 : 12 h light : dark cycle under white light. At the time of this study, the L. danicus Forster strain was discontinued from Penelope Ajani's culture collection, and instead retrieved from the Australian National Algae Culture Collection (ANACC), Hobart, Tasmania; CSIRO strain CS-1164 (Table 1). This strain was maintained in f/2 medium at 20 °C, with ∼60 µmol photons m⁻² s⁻¹ illumination on a 12 : 12 h ligh : dark cycle using LED light panels. All cultures were transferred (1 : 100) into fresh medium at least every three weeks to maintain actively growing cells.

For the experiment, Leptocylindrus danicus strains from each of the four locations were acclimated and grown in locally sourced offshore seawater (0.2 µm filtered and autoclaved) enriched with f/10 medium (Guillard and Ryther, 1962) and modified with the addition of silicate (acidified with 0.5 M HCl in a 1.5 : 1 ratio). Cultures were maintained under controlled light and temperature conditions in Plant Growth Cabinets (CLIMATRON-1100-SL-H, Thermoline Scientific) fitted overhead with LED neutral white light (4000 K). Light was supplied on a 12 : 12 h light : dark cycle, with cultures grown under two light intensities provided at approximately 55 and 100 µmol photons m⁻² s⁻¹ (photosynthetically active radiation; PAR, using a LI-COR Light Meter, LI-250A, equipped with a 2Pi Quantum Sensor). To achieve the lower light intensity, growth cabinets were fitted with neutral density filters (90 % shading; LEE Filters). Cultures were grown at four different temperatures: 16, 18, 20, or 22 °C (Table 2), selected based on the natural average spring sea surface temperature (SST) for the period 2015–2019 at the approximate location of which the strains were isolated (see Table 1). Temperature was kept constant and monitored every 15 min throughout the acclimation and experimental period using iButton^® temperature loggers (Thermochron) (Table 2).

At each light intensity, quadruplicate batch cultures (200 mL) of each strain were acclimated and grown for more than two months prior to measurement. Cultures were inoculated at ∼5000 cells mL⁻¹ into sterile f/10 medium and grown in 250 mL Erlenmeyer flasks at their respective temperatures and irradiances. Throughout the acclimation and experimental periods, all cultures were transferred into fresh media every 10–14 d under sterile (biosafety cabinet) conditions to maintain healthy growing cells. The Twofold Bay L. danicus cultures were unable to establish at 100 µmol photons m⁻² s⁻¹, and therefore, only low light (55 µmol photons m⁻² s⁻¹) incubations were completed for this strain. All measurements were performed on cultures during their exponential growth.

To establish growth rates, aliquots of 1 mL were sub-sampled daily from each flask and fixed with glutaraldehyde (2 % v/v final concentration). Cell counts were performed manually using a Sedgewick Rafter counting chamber by counting to a minimum of 400 cells as per Ajani et al. (2016), using an upright Nikon Eclipse Ci fluorescence microscope (≤200× magnification). Of note, most cultures grew as colonial chains, except for Forster, which was solitary at high irradiance, but formed short chains under low light (Fig. S1 in the Supplement). Growth rates were monitored via light microscopy until the stationary phase was reached. Maximum average growth rates (μMax in divisions per day) for each strain were calculated during the exponential growth phase according to Guillard (1973).

Cell counts and cell volume measurements for data normalisation were taken on the day of experimental sampling for each strain when cultures were in exponential growth using a fixed 2 mL sub-sample (2 % glutaraldehyde). Cells were imaged under a microscope (Nikon Eclipse Ci fluorescence microscope) at 200× magnification using the Infinity Analyze 7 software, (version 7.1.0.1215; Teledyne Lumenera). To estimate cell sizes for each strain and replicate, the first 50 cells encountered were measured using the approximate cell diameters (width) and length, in the image processing package Fiji (Fiji is Just ImageJ) (Schindelin et al., 2012). Cell volume was then calculated assuming cylindrical cells, as per Hillebrand et al. (1999).

To assess biogenic silica (bSi) content and silica incorporation rates, exponentially growing cultures were harvested and inoculated in quadruplicate at 10 000 cells mL⁻¹ into new 250 mL Falcon Tissue Culture Flasks in 155 mL (final volume) of filtered seawater amended with f/10 media. Cultures were inoculated with the fluorescent stain PDMPO (LysoSensor Yellow/Blue DND-160; ThermoFisher Scientific, Australia) to a final concentration of 0.125 µM (Leblanc and Hutchins, 2005) and incubated under experimental conditions for 24 h. Following 24 h incubation, labelled cultures were sampled for cell density, bSi content, PDMPO incorporation, and measurement of the photosynthetic health of cells (FV/FM).

To determine bSi content and silica incorporation rates from PDMPO incubations, 100 mL aliquots from each incubation flask were filtered under low vacuum onto a 47 mm, 0.4 µm Nucleopore Polycarbonate Track-Etch (PCTE) Membrane filter (Cytiva, Whatman). Filters were rinsed with 0.2 µm FSW (1 mL, three times) to remove any residual and unbound PDMPO. Using sterile tweezers, filters were folded into quarters and placed toward the bottom of a 1.8 mL cryovial. Immediately after sampling, samples were flash frozen in liquid N₂ and stored at -80 °C until further processing. Samples were prepared for processing by first thawing and removing filters from cryovials, unfolded onto a filter tower, and covered in 2 mL of 10 % HCl for 2 min to disrupt cell membranes. To further osmotically lyse cells and remove any unbound PDMPO and cellular debris, filters were subsequently rinsed with milli-Q water (1 mL, three times). HCl washed filters were then transferred to 15 mL Falcon tubes to digest frustules in a NaOH-HCl matrix for analyses using established methods (Leblanc and Hutchins, 2005). Briefly, 8 mL of 0.2 M NaOH was added to sample tubes before digestion in a hot water bath (∼95 °C) for 3 h. Samples were cooled in an ice bath and neutralised by adding 2 mL of 0.8 M HCl and mixed well. Filters were removed from sample tubes before centrifugation at 4000 rcf for 5 min to pellet any remaining debris. Samples were then split for processing of bSi and PDMPO analyses.

BSi production in samples were determined by withdrawing 2.5 mL of supernatant for analysis using the colorimetric analysis of reactive silicate, as in Strickland and Parsons (1972). To ensure samples fell within the range of the standard curve, samples (supernatant) were diluted with milli-Q water (2.5 mL final volume) prior to the addition of reagents for bSi content assays on both high light (1 : 10 dilution) and low light (1 : 2 dilution) samples. Absorbance was measured on a UV-VIS Spectrophotometer (Shimadzu UV-1280) at 810 nm and compared against a standard curve (R2≥0.997) prepared with a sodium metasilicate stock (10 mM) to a maximum concentration of 200 µmol L⁻¹. Another 2.5 mL supernatant was withdrawn for measurement of PDMPO in the solution using a spectrofluorometer (Shimadzu RF-6000), equipped with a xenon lamp. Excitation was set at 380 nm (5 nm bandwidth) and emission peaks measured at 542 nm (10 nm bandwidth) with a 200 nm min⁻¹ scanning speed. A standard curve (R2≥0.998) was prepared using a 1/1000 dilution of the original PDMPO stock culture (1 mM) for calibration standards (i.e. 1 µM stock concentration) in the NaOH-HCl digestion matrix to a maximum concentration of 0.1 µmol L⁻¹. All PDMPO data were converted to bSi production using the recommended ratio of 2916 mol bSi per mol PDMPO (McNair et al., 2015) to enable quantitative comparisons. All measurements were made using a quartz cuvette (10 mm light path), results corrected against blanks and normalised to cell density and surface area.

To estimate initial concentrations of cellular uptake and drawdown of phosphate and nitrate/nitrite (NOx) following 24 h PDMPO incubations, filtrate from each replicate was collected from experimental flasks (30 mL) and PDMPO incubation flasks (100 mL) using 0.2 µm PCTE filters. Macronutrients were analysed using the methods of Ringuet et al. (2011) for phosphate assays, and NOx assays based on Schnetger and Lehners (2014). Briefly, phosphate assays were prepared by adding 200 µL of sample to a 96 Well Tissue Culture Plate with 50 µL reducing agent and gently vortexed on a microplate shaker for 30 min. Absorbance was measured at 880 nm and results compared to a standard curve (R2≥0.99) prepared with a potassium phosphate monobasic (KH₂PO₄) stock solution. NOx (NO2-+ NO3-) assays were prepared by adding 180 µL of samples to a 96 Well Tissue Culture Plate with 120 µL reducing agent and incubated at 45 °C for 60 min. Absorbance was measured at 540 nm and results compared to a standard curve (R2≥0.99) prepared with a potassium nitrate (KNO₃) stock solution. All samples were conducted using triplicate technical samples and absorbance measurements made on a TECAN Spark^® Multimode Microplate Reader (Spark Control version 3.1).

The photo-physiological condition of exponentially growing L. danicus cultures were analysed via a multi-step (eight light levels) steady state light curve, using a pulse amplitude modulated (PAM) fluorometer (Water PAM, Walz GmbH, Effeltrich, Germany) following the protocol outlined in Petrou et al. (2011). Prior to analyses, live subsamples (2 mL) were dark-adapted under subdued light for 15 min. Following dark acclimation, a saturating pulse of light for 0.8 s with an intensity of 8 was applied for determination of minimum (FO) and maximum (FM) fluorescence to calculate the maximum quantum efficiency of PSII (FV/FM), where: FV/FM = (FM–FO) /FM. To estimate the efficiency of photosystem II (PSII), a series of saturating pulses (30 s apart) of light under actinic illumination (4 min) at eight intensities for both high light (72, 107, 163, 240, 360, 540, 764, 1050 µmol photons m⁻² s⁻¹) and low light adapted cultures (48, 72, 107, 163, 240, 360, 540, 764 µmol photons m⁻² s⁻¹) were conducted, and the average of the last three measurements at each light step recorded. For each saturating pulse, the effective quantum yield of PSII [Y(II)], non-photochemical quenching (NPQ), and relative electron transport rates (rETR) were calculated automatically by the Water PAM (Walz) in the WinControl-3 software. Using a custom-made macro in SigmaPlot version 14.5 (Systat Software, Inc.), a curve fitting model was applied to rETR as a function of irradiance (PAR; µmol photons m⁻² s⁻¹), and the maximum relative electron transport rate (rETRmax), light utilisation efficiency (alpha; α) and minimum saturating irradiance (Ik) calculated as in Ralph and Gademann (2005). To determine whether PDMPO negatively affected cells, the maximum quantum efficiency of PSII (FV/FM) was determined before and following 24 h incubations.

To determine chlorophyll a and c (c1+c2) content, 50 mL aliquots from each replicate of exponentially growing culture were filtered onto 25 mm Whatman GF/F filters (0.7 µm nominal pore size) under low vacuum (≤15 mm Hg). Filters were placed at the bottom of 15 mL Falcon tubes and immediately stored at -30 °C until further processing. Exposure of chlorophyll extracts to light was avoided to evade degradation of products. Pigments were extracted from cells by adding 3 mL of 90 % acetone (v/v) to thawed samples, vortexed, filters pressed to the bottom of the tube, then allowed to stand in a refrigerator in the dark for at least 15 h at 4 °C. Following overnight extraction, samples were vortexed again, filters removed, then centrifuged at 4000 rcf for 5 min at 4 °C to remove any remaining cellular debris. Chlorophyll content from each sample was determined by loading 2 mL of supernatant into a quartz cuvette (10 mm light path) and absorbance measured at 630 and 664 nm using a UV-VIS spectrophotometer (Shimadzu, UV-1280). Prior to sample measurements, a 90 % acetone blank was run to correct for background fluorescence. Chlorophyll a and c concentrations were calculated (µg mL⁻¹) using the equations of Jeffrey and Humphrey (1975), later refined by Ritchie (2006), then normalised to cell density and cell volume.

For POC/PON determination, samples (100 mL) of exponentially growing cultures (one per biological replicate) were collected by vacuum filtration (<20 mm Hg) onto pre-combusted (450 °C for 5 h) 47 mm Whatman GF/F filters (0.7 µm nominal pore size). Individual filters were then wrapped in pre-combusted aluminium foil and immediately stored at -30 °C until further processing. Filters were prepared for analysis by first thawing, then acidifying with 1 M HCl dropwise until the filter was saturated. Acidified filters were then left in a drying oven at 60 °C for at least 48 h before being cut into quarters, wrapped and pelletised in 9 mm tin capsules (Elemental Microanlysis, UK). To quantify and correct for background concentrations, four “blanks” (filters with 100 mL of f/10 medium only) were also collected onto pre-combusted GF/F filters and processed, as above. Carbon and nitrogen 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.

All figures and data analyses were performed in RStudio version 2024.12.0+467 (R Core Team, 2024). Morphological and physiological trait data were tested for temperature-driven responses using linear regression. When no significant linear relationship was found, a one-way analysis of variance (ANOVA) was used to test for significant differences in phenotypic traits between strains at each light intensity. To ensure assumptions of normality and equal variance were satisfied, a Levene's test for homogeneity of variance was applied to all data, a priori. When treatment groups varied significantly, a Tukey's honestly significant difference (HSD) post hoc was used to isolate which groups differed significantly from each other. In cases where the assumption of equal variances was violated, a Bartlett's test was performed, followed by a Welch's ANOVA. If the F ratio was found to be significant, a Games-Howell post hoc test was performed to determine which treatment groups differed. Statistical significance was determined at α=0.05 for all samples. All figures were generated using add-on package ggplot2 v. 3.5.1 (Wickham, 2016).

Cell size is considered a key phenotypic trait influencing the extent at which phytoplankton respond to changes in the environment (Litchman et al., 2010). This study revealed significant intraspecific morphological variability between L. danicus isolates in response to acclimation to their locally adapted thermal niche (as experienced during spring), including differences in chain forming and chain length between strains. We confirmed a broad thermal niche in L. danicus, as in Ajani et al. (2021), and uncovered a strong temperature-dependent and latitudinal effect on growth rate and individual cell size (surface area and volume) for each strain, independent of whether strains were colonial or solitary. Under low light, low temperature conditions (Twofold Bay), we saw a significant reduction in mean growth rate which was anticorrelated with higher cell surface area to volume ratio. This higher surface area to volume ratio, typical of smaller cells, provides a competitive advantage over their larger counterparts, such as improving nutrient uptake through the reduced diffusive boundary layer thickness, and lower nutrient requirements to achieve maximum growth (Aksnes and Egge, 1991). Generally, phytoplankton are expected to decrease in size with increased surface warming, due to resource constraints and increased division rates (Hattich et al., 2024; Sommer et al., 2017). The strength of these shifts toward smaller species are likely to intensify with longer and more intense MHWs, and reduced nutrient supply from weakened coastal upwelling (Zhan et al., 2024). However, the response of diatoms to increasing temperatures is conflicting, with some species increasing in cell size, likely due to evolving different optimal growth temperatures (Liang et al., 2019). The increase in cell size of L. danicus at higher temperatures may be a direct physiological response to warming due to differences in phenotypic thermal performance (Baker et al., 2016; Schulte et al., 2011; Sheehan et al., 2020) or driven by an increased need for resources (Hattich et al., 2024).

Climate-induced changes in phytoplankton physiology and community structure are likely to alter the efficiency of the biological carbon pump (Basu and Mackey, 2018). This is because the nutritional quality (elemental and biochemical composition) and quantity of organic matter that ultimately influences the transfer of energy to higher trophic levels and carbon export to depth is largely dependent on species composition, their abundance, and their individual elemental stoichiometry (Duncan et al., 2022; Finkel et al., 2010). We found that under low growth irradiance, Coffs Harbour strains (22 °C) reduced their chlorophyll a content per unit area, compared to the other strains that displayed increased intracellular chlorophyll a with decreasing temperature. We also saw that low light acclimated Coffs Harbour strains had enhanced photosynthetic capacity through higher rETR and lower NPQ. This was also reflected in the increased rETRmax and Ik. The highest rates of NPQ were seen in the low-light acclimated Maroubra strain, which correlated with lower photosynthetic efficiencies, suggesting light limitation. The relatively lower intracellular chlorophyll a content of the larger, low light grown Coffs Harbour cells can be explained by the limitation of self-shading of pigments within the cell due to packaging effects with cell size, therefore providing absorptive efficiency of photosynthetic pigments (Finkel et al., 2004).

The low C : N ratios shown by Coffs Harbour strains grown at low light are consistent with a previous study that found light limitation caused a change in the uptake ratio and elemental composition of diatoms, specifically reduced carbon (Saito and Tsuda, 2003). In a warmer ocean, phytoplankton access to nutrients are proposed to decline, altering elemental ratios, while simultaneouly access to light for cells entrained in the surface layer will increase, reducing carbon limitation. In this study, the differences in elemental composition were not correlated to temperature or growth rates, but shifted in response to growth irradiance in the Coffs Harbour strain, suggesting that the stoichiometric plasticity is a response to low light acclimation (Sauterey and Ward, 2022; Yang et al., 2020). Therefore, under anticipated conditions caused by climate change induced stratification, our results indicate increased C : N with higher light availability could lead to decreased food quality (lower nitrogen supply) for grazing zooplankton (Finkel et al., 2010; Marinov et al., 2010). Importantly however, the strains used in this study were grown under nutrient replete conditions, thus precluding nutrient limitation as an influencing factor (Table S2), therefore the inclusion of nutrient manipulations in future temperature and light experiments would be important to further advance knowledge on how species adapt under the triad (temperature, light and nutrients) of critical environmental factors.

Diatom size and silicification are both important phenotypic traits driving species performance, in turn influencing the structure and functioning of diatom assemblages and food web dynamics (Finkel et al., 2010; Sommer et al., 2017). Together, growth morphology and silica content, which varies across diatom species and strains (Martin-Jézéquel et al., 2000), can influence susceptibility to grazing pressure and sinking capacity (Tréguer et al., 2018). Our study revealed a clear latitudinal trend in silicification with temperature, and revealed strong size-dependent relationships, independent of whether chain-forming or solitary. We observed a reduction in silica deposition per cell with lower temperature (under low growth irradiance), however when differences in individual cell surface areas were accounted for, there was little difference in silicification rates across strains. When normalised to cell volume, strong trends appeared with lower temperature strains showing higher incorporation rates. Together, these results infer that under low light conditions, L. danicus from Twofold Bay (16 °C) have thicker frustules, which may in part be due to the lower growth rates of this strain, and therefore the increased length of time cultures underwent the cell wall synthesis phase, i.e. allowed for maximum silica incorporation into the frustule (Martin-Jézéquel et al., 2000; Su et al., 2018). The higher silica content in these cells would result in higher density and therefore more efficient sinking, as well as increased mechanical protection (in response to copepod cues), which are key ecological strategies for the smaller, slower growing diatoms that may otherwise be easily grazed in surface waters (Grønning and Kiørboe, 2020; Liu et al., 2016).

On the other hand, under high light conditions, L. danicus showed the highest silica incorporation per cell, per surface area and per volume under the lowest temperature (18 °C, Maroubra strain), confirmed by the highest bSi : C ratio, contrasting with the typical inverse correlation of silicification to growth rate (Flynn and Martin-Jézéquel, 2000). This strong increase in silicification of strains grown at lower temperature is consistent with previous studies looking at temperature-regulated responses on silicification using the diatom T. pseudonana (Baker et al., 2016; Sheehan et al., 2020). These data suggest that while the mechanistic uptake of silica, polymerisation and deposition of frustules may be thermally constrained, and independent of cell division (Sheehan et al., 2020), a combination of both increased temperature and light may elicit a similar diminished rate of silica deposition as seen in other stressors, such as seawater acidification, as seen in some species of an Antarctic diatom community (Petrou et al., 2019), high irradiance and pCO₂ in T. weissflogii (FitzGerald-Lowry et al., 2025), or osmotic stress at higher salinities (Vrieling et al., 1999). Considering the phenotypic differences observed here, in the context of future oceans with combined warmer temperature and increased pH where diatoms are larger with reduced silica content, these data would suggest that diatoms in coastal upwelling systems may be subjected to increased grazing pressure, making them less efficient vectors for silica export (via sinking), and influencing trophic energy transfer to higher levels (Tréguer et al., 2018; Zhang et al., 2017).

As temperatures in the EAC continue to warm, these data suggest that we may see changes to primary productivity and a reduction in silica production rates in L. danicus in this region. Such responses could have important implications for local silicon cycling and carbon sequestration given the dominant role of L. danicus in coastal communities and spring blooms on the Australian coast (Ajani et al., 2016). The large strain variability observed for L. danicus, a non-isolated cosmopolitan species, indicate that strain-specific responses to temperature are unlikely to be uniform (Ajani et al., 2021; Ajani et al., 2016; Koester et al., 2013). Therefore, further studies are necessary to isolate whether such responses are solely due to strain variability or whether the responses we are seeing are in fact temperature-driven (owing to their phenotypic plasticity). While L. danicus is known to encompass a number of functional traits that allow them to rapidly diversify and adapt with fluctuating conditions in upwelling systems (Ajani et al., 2021; Ajani et al., 2016), the potential reductions in cell density under warmer conditions may counteract the success of this species in this region. As such, further studies into species-specific responses with temperature are required to provide a better understanding of the major taxonomic groups that are likely to be affected under future oceanic conditions. This will enhance our understanding of how warmer oceans will alter species composition, abundance, distribution and food web interactions, which are essential for shaping marine nutrient cycling and carbon export.