Quantifying diatom silicification with the fluorescent dye, PDMPO

Abstract

Diatoms require silicic acid to construct ornately detailed cell walls called frustules. The growth and geographic distribution of diatoms is often controlled by the availability of silicic acid. Analytical methods exist to assess diatom community biogenic silica (bSiO₂) production, but partitioning production among taxa has been largely qualitative. We present a method for the quantitative analysis of taxa-specific silica production through labeling diatoms with the fluorescent dye PDMPO [2-(4-pyridyl)-5-((4-(2-dimethylaminoethylaminocarbamoyl)methoxy)phenyl)oxazole]. To make PDMPO a quantitative tool: diatom frustules were solubilized to assess the total diatom community incorporation by quantitation of PDMPO fluorescence using a fluorometer, and laser confocal microscopy was used to quantify the fluorescence of PDMPO in single diatom cells. We created a fluorescence standard to intercalibrate the raw fluorescence signals of the fluorometer and microscope and to determine the fluorescence per mole of PDMPO. PDMPO incorporation was converted to silica production using diatom bSiO₂:PDMPO incorporation ratios which varied systematically with silicic acid concentration. Above 3 μM Si(OH)₄, bSiO₂:PDMPO was constant and PDMPO incorporation was converted to silica production using a mole ratio of 2,916 as determined from cultures. Below 3 μM, the ratio was a linear function of [Si(OH)₄] (bSiO₂:PDMPO = 912.6 × [Si(OH)₄]), as determined using data from two oceanographic cruises. Field evaluation of the method showed that total community PDMPO incorporation generally agreed to within 30% of radioisotope-determined silica production. This PDMPO method has the potential to be a powerful tool for understanding physiology, silicification and resource competition among diatom taxa.

Introduction

Diatoms are one of the most ecologically diverse and successful organisms in the ocean; they account for 20 – 40% of total oceanic primary production (Nelson et al. 1995), play a fundamental role in a number of global biogeochemical cycles and serve as the base of many productive aquatic food webs (Mann 1993). High productivity, large cell size and mineral-ballasted cell walls make diatoms an important vector for carbon export from the surface ocean (Alldredge and Silver 1988; Buesseler 1998). Diatoms are unique among phytoplankton due to an obligate need for silicon to produce cell walls of polymerized silica known as frustules. Diatom growth and productivity can therefore be strongly influenced by the availability of silicic acid, i.e. dissolved silica (Brzezinski 1992). Despite silicon being the second most abundant element in Earth's crust, diatom consumption of silicic acid depletes concentrations in the surface-ocean to levels where diatom Si uptake (Brzezinski et al. 2008; Krause et al. 2012) and growth rate (Dugdale et al. 1995) can become Si limited.

Silicic acid availability can alter the relative abundance of diatoms within phytoplankton assemblages (Tilman 1982). Such composition shifts can drive changes in the efficiency of the biological pump where size and growth of phytoplankton are linked to sinking and export from the surface ocean (Martin et al. 1991; Harrison 2000). The large cell size of diatoms, relative to smaller phytoplankton, results in the more efficient transfer of energy to top trophic levels, supporting larger macrozooplankton biomass than smaller phytoplankton taxa (e.g. Pomeroy 1974). Understanding the distribution of diatom silica production and how silicic acid concentration affects diatom community structure is key to understanding and modeling energy flow through aquatic food webs.

Currently, the analytical methods to measure rates of Si use by diatoms in natural communities include measuring net changes in silicic acid (Pondaven et al. 2000) and biogenic silica (bSiO₂) (Brzezinski and Nelson 1989; Krause et al. 2010) over time, or determination of gross rates of bSiO₂ production using silicon isotope tracers (Goering et al. 1973; Nelson et al. 1976; Brzezinski et al. 1997). To better understand the influence of silicic acid availability on natural diatom assemblages, knowledge of interspecies variability in silicic acid uptake and species-specific contributions to bSiO₂ production is required. Previous work suggests that Si limitation can shift diatom species composition, where low Si environments select for lightly silicified species with lower Si requirements (Dortch et al. 2001). Diatom community composition, dictated by nutrient availability and grazing pressure, can play a substantial role in community silicification (Assmy et al. 2013). The elemental stoichiometry of diatoms can differ depending on nutrient availability and can cause a preferential drawdown of one macronutrient over another (Brzezinski et al. 2002). Nutrient availability may also alter the degree of diatom silicification: frustule thinning occurring in low silicic acid conditions (Brzezinski et al. 2011a) and thickening in low iron conditions (Hutchins and Bruland 1998).

A number of attempts have been made to partition total biogenic silica biomass or total silica production among the different diatom groups present within natural diatom assemblages. Blain et al. (1997) used numerical abundance and surface area to estimate the contribution of different species to total biogenic silica production in the equatorial Pacific. This method assumed a constant bSiO₂ per unit surface area for estimating the bSiO₂ production for all species, and assumed that all cells were active and growing at the same rate. However, frustule thickness can vary at least five-fold among species grown in nutrient replete conditions (Paasche 1980) and three-fold for cells of the same species under nutrient stress (Brzezinski et al. 1990). Identification of actively silicifying cells was addressed by Shipe and Brzezinski (1999) who used autoradiography with the radioisotope ³²Si to visualize newly formed frustules in Rhizosolenia mats from the North Pacific subtropical gyre. With the autoradiograph, the fraction of active cells and the doubling time of species could be measured in the mixed assemblages of the mats. Difficulties with this method are the cost of ³²Si (∼$1200 per μCi) and the considerable lag time between taking and analyzing a sample due to the need for ³²Si to reach equilibrium and the months of exposure to create the autoradiograph. An alternative probe was needed, something that would label newly deposited silica, was inexpensive, and quicker to visualize. Brzezinski and Conley (1994) found that the fluorescent dye, Rhodamine 123, was incorporated into newly deposited silica when present in growth media and could trace silica deposition in diatom cultures with negligible toxic effects. However, the low fluorescent yield, low incorporation ratio of the dye with bSiO₂, and emission overlap of Rhodamine 123 with chlorophyll made the dye difficult to use. More recently, a different fluorescent dye, 2-(4-pyridyl)-5-((4-(2-dimethylaminoethylaminocarbamoyl)methoxy)phenyl)oxazole (PDMPO) has proved more useful for tracking newly deposited biogenic silica in frustules (Shimizu et al. 2001).

PDMPO was first synthesized by Diwu et al. (1999) as a pH indicator for acidic cellular organelles. Shimizu et al. (2001) found that like Rhodamine 123, PDMPO was incorporated into the silica deposition vesicle (SDV) and ultimately into newly formed diatom frustules, and that the fluorescence intensity increased when the dye was in the presence of silicic acid. LeBlanc and Hutchins (2005) developed a method to make PDMPO a useful tool for identification of cells actively depositing bSiO₂ within a natural assemblage. One of the key findings from LeBlanc and Hutchins's (2005) work was evidence that PDMPO was incorporated in a nearly constant ratio with biogenic silica. Thus, quantitative silicification rates of the total diatom community can be measured by applying the bSiO₂:PDMPO ratio to the amount of PDMPO incorporated (LeBlanc and Hutchins 2005). However, determination of silica production in single cells by PDMPO incorporation was still qualitative (Leblanc and Hutchins 2005). Znachor and Nedoma (2008) developed a method to image and measure single cell PDMPO fluorescence to look at relative differences in diatom species silicification, but estimates of the absolute amount of Si incorporation were not reported. Since these studies, PDMPO has been employed in a number of diatom ecology studies (e.g. Durkin et al. 2012, 2013; Saxton et al. 2012; Znachor et al. 2013), but none have used PDMPO as a quantitative tool for allocating Si incorporation among diatoms in a natural assemblage.

Several challenges must be overcome to fully realize PDMPO as a quantitative tool for studying bSiO₂ production in natural diatom assemblages. To measure the total community silicification with PMDPO by fluorometry, diatom silica must be solubilized without degrading the incorporated dye. Dissolving the labeled frustules eliminated the need to quantify how many cells are in the excitation and emission views of the fluorometer and it avoids complications due to heterogeneous particle distribution. Using PDMPO to measure silica production at the single cell level requires quantification of single cell fluorescence and accurate calibration of the relationship between incorporated PDMPO and the amount of new silica polymerized. Because cells are solubilized for fluorometry but visualized intact using microscopy, intercalibration between instruments requires a fluorescence standard which can be used in both measurement states. In this paper we describe improvements to the PDMPO method that take PDMPO labeling from a qualitative assessment to a quantitative measurement of newly deposited silica in both natural diatom assemblages and in individual diatom cells.

Materials and Procedures

Fluorescent dye

The fluorescent dye, PDMPO, LysoSensor Yellow/Blue DND-160, was used to label newly deposited diatom frustules. Vials of 1 mM PDMPO stocks (50 μL volume in dimethylsulfoxide) were purchased from Molecular Probes and stored in the dark at -20° C until use. PDMPO has an excitation maximum within UV wavelengths (357 – 377 nm) and has dual emission maxima, 417 – 483 nm and 490 – 530 nm in the pH range of 5.0 – 7.0 (Diwu et al. 1999). However, in acidic conditions, or in the presence of hydrated amorphous silica, the dual emission peaks collapse into a single peak at 534 nm (Shimizu et al. 2001).

Labeling Protocol

A vial of PDMPO was thawed and centrifuged to consolidate the dye into a single droplet and facilitate quantitative transfer to experimental samples. Previous work found that incubations with the manufacture's recommended 1 μM concentration decreased growth rates in longer incubations (48 and 96 hours), but this apparent toxicity could be overcome by decreasing the dye concentration to 0.125 μM, which still provided adequate and consistent labeling (LeBlanc and Hutchins 2005). The protocol, described here, results in a final PDMPO concentration of 0.157 μM in seawater samples (48 μL PDMPO in 306 mL sweater). This concentration produced well-labeled cells and did not affect growth rate in 8 species tested (t-test, p=0.36, 10 experiments total among 8 species).

Laboratory cultures were incubated for 24 hours and field samples were incubated for 4 – 24 hours, depending on the goals of each experiment. Cultures were incubated at 50 μE m^-2 s^-1 under cool fluorescent light (14/10 h light dark cycle) at 16° C. In field samples, PDMPO was added to 250 mL polycarbonate bottles (brim volume 306 mL). The bottles were then placed in light bags with various combinations of mesh screening to mimic the light intensity at the depths of sample collection and incubated in deckboard incubators cooled by flowing surface seawater. After incubation, a 100 mL aliquot of the labeled sample was removed from the bottle and reserved for measurement of single cell PDMPO incorporation; the rest of the sample was used to determine total community PDMPO incorporation. All culture samples and most natural phytoplankton samples were incubated with a paired replicate sample that did not contain PDMPO in order to quantify background fluorescence for the total community incorporation measurements. An additional separate 100 mL sample was collected at the beginning of the incubations and fixed with 4 mL of Bouin's solution for cell identification and enumeration.

Total Community PDMPO incorporation

To quantify total community PDMPO incorporation unbound PDMPO was removed from the sample and the diatoms were dissolved for fluorometric analysis as follows. Samples were filtered onto polycarbonate filters (1.2 μm pore, 25 mm diameter). The cells and filter were transferred to a 15 mL polypropylene conical centrifuge tube, covered in 10 mL of 100% methanol (American Chemical Society grade) and placed in the dark at 4° C for 24 hours to remove unbound PDMPO and to eliminate photopigments. This methanol extraction was tested on diatom cultures and largely eliminated the unbound PDMPO (Figure 1) and most of the photopigments, thereby lowering the background fluorescence and increasing measurement sensitivity. After extraction, the filter was compressed to the bottom of the tube with a Teflon rod and centrifuged (10 min, 1230 × g) to pellet the cells. The methanol supernatant was aspirated to 1 mL, ensuring all cells remained in the sample. Test tube, filter and remaining methanol were dried, uncapped, in a warm (<60° C) vacuum oven until the methanol evaporated, leaving the sample dry. The effect of drying on PDMPO fluorescence was evaluated by comparing known quantities of PDMPO in 1 mL of methanol that were either dried at room temperature or at 60° C. There was no statistically significant difference between the fluorescence of the samples dried at the two temperatures (p=0.31, t-test, n=10).

Figure 1.

Comparison of rinse methods (0.2 μm filtered sea water, 10% HCL, and 100% Methanol) to eliminate unbound PDMPO. The black bars represent fluorescence in T. weissflogii and the gray bars represent fluorescence of C. socialis.

Frustule-bound PDMPO in the dried samples was solubilized and quantified fluorometrically. When dry, 0.2 mL of 0.5M HF was added to the crushed filter and cells and all contents were mixed with a Teflon rod to remove air bubbles from the filter to allow contact between cells and the HF. After a 1-hr digestion, the samples were neutralized with 2.8 mL of saturated boric acid (∼1 M). The fluorescence of the solution was quantified using a Trilogy Laboratory Fluorometer (Turner designs) with the crude oil snap-in module (Light Emitting Diode (Center Wavelength) 365 nm, excitation 350/80 nm, emission 410 – 600 nm). Raw fluorescence was converted to PDMPO concentration using a standard curve of known concentrations of PDMPO in the same chemical matrix (HF/boric acid). Hot NaOH digestion was found to degrade the fluorescence of the dye (Figure 2) and therefore, dissolution with HF is recommended over dissolution with sodium hydroxide in a hot water bath (95°C) as is used in determining biogenic silica concentration (Paasche 1973). No degradation of dye fluorescence was observed with the HF dissolution. Once boric acid is added, the fluorescence is stable for at least one month (data not shown).

Figure 2.

Standard curves of PDMPO in 0.2 N NaOH. Open circles represent standards that were not heated. Squares represent the fluorescence of the same standards after one hour at 98°C.

Paired seawater blanks without PDMPO were processed the same way as the labeled samples and allowed for correction of signal from autofluorescence. Blanks from field samples scaled with chlorophyll concentration (R²=0.39). For field samples where we did not have a paired blank, a linear regression was derived from the relationship between PDMPO and Chl a was used to estimate the blank for the sample (Equation 1).

$$\mathit{\text{Blank}}\left[\text{PDMPO}\right]=0.005\times \left[\text{total Chl}\right]+0.04$$ (1)

where Blank [PDMPO] is the equivalent concentration of PDMPO of the blank in nM units and [total Chl] is the total chlorophyll concentration (μg L^-1) in the seawater as measured by fluorometry (Smith et al. 1981). The uncertainty in estimated blank values (∼0.1 nM PDMPO) is 10 – 50 times smaller than the typical sample reading.

Single Cell PDMPO incorporation

Quantitative assessment of single-cell PDMPO incorporation requires accurate measurement of the quantity of dye incorporated into the frustule of a cell. Cells were mounted on slides and imaged using laser confocal microscopy to quantify the fluorescence from PDMPO within single diatom cells. Excitation of the dye with a laser offered more consistent energy than a mercury lamp and the precise depth imaging of the confocal microscope offered finely-resolved three-dimensional fluorescence reconstruction. The raw fluorescence signal collected from the confocal was converted to an absolute amount of PDMPO incorporated using a fluorescence standard (see below).

Following experimental incubations the seawater subsample for microscopy was immediately centrifuged for ten minutes (1230 × g) and the supernatant removed. As with the total community incorporation measurement, unbound PDMPO and most photopigments were removed by the addition of 10 mL of methanol and storage for at least 24-hrs in the dark at 4° C. Samples can be stored in this state for up to 2 years (the longest period tested). Slides were prepared by resuspending an aliquot (20,000 to 40,000 cells based on enumeration from an independent Bouin's preserved sample) of the methanol and cell mixture in 10 mL of 0.2 μm filtered seawater and then filtering the suspension under low-vacuum (<13 kPa) onto a polycarbonate filter (1.2 μm, 25 mm). The filtered was placed, sample-side down, on a drop of deionized distilled water (>18 MΩ) on a polylysine-coated glass slide. Flash freeze spray was applied to the opposite side of the slide from the sample. The filter was then quickly peeled back and removed using forceps, leaving the cells on the slide without the filter (Franck 2002). After the slide dried at room temperature in the dark, a cover slip was applied with ProLong gold antifade (Life Technologies). The coverslip was sealed to the slide with clear nail polish and stored in the dark at 4° C.

Cells were imaged using an Olympus Fluoview 1000 Spectral Confocal microscope at the Neuroscience Research Institute at University of California Santa Barbara and using a Nikon A1 Confocal Microscope at the University of South Alabama. Some photopigments that fluoresce in the same excitation/emission settings as PDMPO remained in cells after extraction in methanol (Figure 3, top panel). That interference was removed by imaging in two wavelength channels (Figure 3, middle panel), one configured to best match the wavelength excitation/emission for PDMPO and the other employing a longer wavelength excitation/emission combination configured for photopigments. The PDMPO channel was excited with a 405 nm diode laser and emissions measured from 450 – 550 nm. Photopigments were excited with a 559 nm diode-pumped laser and emissions measured from 618 – 718 nm. Images were acquired in sequence, first exciting with the 405 nm laser then the 559 nm. Cell fluorescence was imaged in three dimensions using an optimized z-step, which is half the axial resolution of the objective: 420 nm on the Olympus confocal and 250 nm on the Nikon confocal.

Figure 3.

False-color images of Pseudo nitzschia sp. cells labeled with PDMPO. Top view is the total fluorescence of a chain of two cells in the PDMPO channel. The middle view is the PDMPO channel, emission from 450 – 550 nm (blue) overlaid with the longer wavelength 618 – 718 nm photopigment channel (red). Bottom view is the corrected PDMPO fluorescence.

Images were analyzed using Imaris 7.6.5 software (Bitplane). Total cellular fluorescence was calculated using the surface function in Imaris. A surface was created around voxels above a threshold fluorescence, essentially creating a three dimensional volume with ∼420 nm or ∼250 nm depth resolution. Voxels that fluoresced in the long wavelength channel were located and removed from the short wavelength image to create a surface that represented only PDMPO fluorescence (Figure 3, bottom panel; see Assessment section). Total PDMPO fluorescence was calculated by summing fluorescence intensity values of all voxels within the photopigment-corrected volume.

Fluorescence Standard

Standardized fluorescence (RFU _confocal) values for individual cells from the confocal microscope were converted to the standardized fluorescence of the fluorometer (RFU _fluorometer) and then to absolute amounts of PDMPO (moles). This required determination of the RFU _confocal: RFU _fluorometer ratio which was accomplished using a fluorescence standard of uniformly labelled Thalassiosira weissflogii (CCMP 1051). Alternative fluorescence standards proved insufficient. Fluorescent beads were applicable to the microscope, but were not an appropriate for the fluorometer because the volume illuminated in the cuvette on the fluorometer was unknown and beads can clump or sink (albeit slowly), these factors hampered calculation of fluorescence yields of a single bead on the fluorometer. Solutions of PDMPO could be analyzed on the fluorometer but proved problematic on the microscope. A labelled diatom met the requirements for a standard that could be analyzed on both the fluorometer and on the confocal microscope to intercalibrate the two instruments.

The identity of the diatom used to construct the standard is immaterial. Most important is that the number of cells analyzed is known and that the cells are uniformly labelled with PDMPO. This allows the average cellular fluorescence for individual cells determined by the microscope to be compared directly to the average cellular fluorescence determined by fluorometry (RFU_fluorometer ÷ number of cells dissolved). The quotient of the cellular fluorescence from the two instruments yields the RFU _confocal: RFU _fluorometer which quantifies the difference in the fluorescence representing a fixed mass of PDMPO for the two approaches.

The fluorescence standard was prepared by growing T. weissflogii in batch culture, in f/2 media with 300 μM silicic acid. T. weissflogii was an ideal species for the standard because it does not form chains, making individual cell measurements easier, and it is easy to grow. The high concentration of silicic acid avoided cell limitation of Si uptake by low silicic acid which can lower cellular silicon content (see Martin-Jézéquel et al. 2000). This resulted in even labeling (standard error of individual cell fluorescence on the confocal was 5% of the average) of the frustules as PDMPO was incorporated over multiple generations under Si-replete growth conditions (Brzezinski and Conley 1994). The experimental culture was first acclimated in media without PDMPO (three doublings), then a subsample was transferred to media containing 0.148 μM PDMPO with high-silicic acid. The culture was harvested after five doublings, when 96% of the cells had fully labeled frustules and processed as described above.

Fluorescence normalization on the confocal microscope

Fluorescence intensity varied significantly among diatom taxa, and within diatom taxa, consistent with the results of Durkin et al. (2012, 2013). This made it necessary to adjust the photomultiplier voltage on the Olympus confocal and the gain on the Nikon confocal to ensure measureable PDMPO fluorescence signal without saturating the detector. Readings taken at different voltages were normalized to 650 V on the Olympus confocal and a gain of 4 on the Nikon confocal, as these were the most frequent settings employed when measuring samples. The relationship of fluorescence as a function of voltage or gain was obtained by measuring uniformly labeled cells of T. weissflogii that also served as the fluorescence standard described above, at settings that spanned the range of values used to image all cells. This produced an exponential relationship (R²=0.92) between voltage and PDMPO fluorescence on the Olympus confocal and a linear relationship (R²=0.81) between voltage and PDMPO fluorescence on the Nikon confocal.

Measurements of single cell PDMPO incorporation on the two different microscopes produced very similar estimates of total PDMPO incorporation, within the variation of the sample. Data from the Olympus microscope at University of California Santa Barbara was compared with the Nikon microscope at University of South Alabama by imaging the T. weissflogii standard on each instrument in addition to one sample from a recent cruise. The T. weissflogii standard was used to convert the cruise sample from measured RFU to moles of PDMPO yielding a value of 0.22 ± 0.22 (SD) nmol PDMPO cell^-1 for the Olympus confocal and 0.25 ± 0.12 (SD) nmol PDMPO cell^-1 for the Nikon.

bSiO₂:PDMPO incorporation ratio

The methods described above detail how to quantitatively measure total PDMPO incorporation by a diatom culture, a natural diatom assemblage or a single diatom cell. PDMPO incorporation can be converted to Si production using knowledge of the bSiO₂:PDMPO ratio in diatoms provided this ratio is reasonably consistent among diatom taxa. Work by LeBlanc and Hutchins (2005) with two cultured species support this assumption; the standard error of the bSiO₂:PDMPO ratio was 24% of the mean. The present method expands on their work to further examine the central tendency and variability of the ratio using the protocols described above. Culture experiments were conducted with actively growing cells (growth rates from 0.2 to 0.8 day^-1) to measure change in biogenic silica and the incorporation of PDMPO over 24 hours (Table 1). Culture species were a combination of isolates established from the California coast and cultures purchased from the National Center for Marine Algae and Microbiota (East Boothbay, Maine USA).

Table 1.

Diatom cultures used to determine mol:mol ratio of Si deposited to PDMPO incorporated. SC77 was isolated by Holly Bowers

Thalassiosira weissflogii (CCMP 1051, isolated 5/13/1985)

Chaetoceros socialis (CCMP 172)

Odontella aurita (CCMP 595)

Corethron hystrix (CCMP 308)

Pseudo-nitzschia multiseries (SC77, Huntington Beach)

Isolated from California coast July 2012

Chaetoceros socialis (Avila Beach)

Melosira varians (Santa Cruz)

Chaetoceros didymus (Monterey Bay)

Cultures were incubated in pairs, one control (blank) without added PDMPO and one with added PDMPO, at ∼16° C for 24 hours (light/dark cycle of 14/10). Biogenic silica concentration was measured at the beginning and end of each experiment using a NaOH digestion in teflon tubes (Krause et al. 2009) and colorimetric ammonium molybdate method (e.g. Brzezinski and Nelson 1995). PDMPO incorporation was measured using the fluorometer in parallel as described above. The mole ratio of the increase in biogenic silica to the incorporation of PDMPO (bSiO₂:PDMPO) for eight species had a median value of 2,916 ± 708 (SE), n=8. This incorporation ratio was similar to what LeBlanc and Hutchins (2005) found in their culture experiments where the average ratio was 2,800 ± 780, despite the differences in measurement protocols. We use and report the median value instead of the mean for the culture data because the frequency distribution of the data skew towards higher values which biases the average.

A summary of the major steps of the method described above are presented in Figure 4.

Figure 4.

Flow chart for quantitative use of PDMPO for measuring diatom silicification. Not shown is the separate sample required for enumeration of diatom abundance and assemblage structure.

Assessment

Field assessment of the bSiO₂: PDMPO incorporation ratio

Field measurements of biogenic silica production and PDMPO incorporation were used to verify the applicability of the laboratory-derived bSiO₂:PDMPO ratio during two scientific cruises. Sampling occurred during the “Dye labeling of diatoms” (DYEatom) cruise from June 27 to July 5, 2013 aboard the R/V Point Sur off the California coast between Monterey Bay and offshore of Bodega Bay and on the Iron Bruland (IrnBru) cruise from July 4 to July 21, 2014 aboard the R/V Melville off the west coast of the United states between San Diego, California and Southern Oregon. Total community PDMPO incorporation was compared with total community silica production measured with the radioisotope ³²Si. All incubation volumes (306 mL) were consistent between the two methods, PDMPO and ³²Si, and across samples. Silica production samples were spiked with 262.8 Bq of high-specific activity ³²Si (15,567 Bq μg^-1 Si). At the end of each incubation, samples were filtered through 1.2 μm polycarbonate filters and ³²Si incorporation determined as in Krause et al. (2011). Total community PDMPO incorporation was measured following the protocol described above using the same seawater samples and type of filters as used to measure ³²Si incorporation.

The bSiO₂:PDMPO mole ratio from the cruise samples was calculated as the slope of the regression of ³²Si incorporation rates and PDMPO incorporation rates using a Model II, reduced major axis regression (Bohonak 2004). The resulting average bSiO₂:PDMPO mole ratio was 2,454 ± 89 (SE) (n = 201, R²=0.74) with a 95% confidence interval (2,279 to 2,629) (Figure 5). Previous work suggested that the bSiO₂:PDMPO incorporation ratio may depend on the concentration of silicic acid in the environment (Durkin et al. 2013). We observed that the ratio did not show a significant trend with silicic acid concentrations over 3 μM (R²=0.01) (Figure 6a); however, below 3 μM the ratio decreased linearly with respect to the concentration of silicic acid (R²=0.64) (Figure 6b). Changes in the bSiO₂:PDMPO incorporation ratio with respect to silicic acid concentration have been seen in other studies as well (Jennifer Long, personal communications). It is recommended that the bSiO₂:PDMPO mole ratio be calculated from the regression as 912.6 × [Si(OH)₄] when ambient silicic acid concentrations are below 3μM. Note that the regression is forced through the origin based on the assumption that silica production and PDMPO incorporation cease at a [Si(OH)₄] of zero.

Figure 5.

Field assessment of the bSiO₂: PDMPO incorporation mole in a Model II regression between PDMPO incorporation and new biogenic silica production measured from ³²Si from two different cruises. Closed gray circles are data points from the DYEatom cruise, open circles are data from IrnBru cruise.

Figure 6.

bSiO₂:PDMPO mole ratio from field samples plotted across the range of silicic acid concentrations observed on two cruises. Closed gray circles are data points from the DYEatom cruise, open circles are data from IrnBru cruise. (A) The regression line was fit to all data above 3 μM [Si(OH)₄]. (B) Enlarged plot of the boxed section from (A) fit with a linear regression forced through the origin.

Figure 7 shows the result of a second approach to finding the central tendency of the bSiO₂:PDMPO mole ratio from the cruise samples when [Si(OH)4] exceeds 3μM. The cubed root of the ratio was taken to normalize the data. The median of this transformed data corresponds to a bSiO₂:PDMPO mole ratio of 3,402. The slope of the regression and the median of the transformed data are within 15% and 17% respectively, of the ratio from the diatom cultures. Although there is significant variability in the bSiO₂:PDMPO incorporation ratio, the variance is such that using the median culture ratio to calculate silica production from total community PDMPO incorporation for field samples where [Si(OH)₄] exceeds 3μM will yield production rates that are within 30% of rates determined using ³²Si. In cases where ambient [Si(OH)₄] is < 3 μM using the recommended regression produces agreement within 36%.

Figure 7.

The normally transformed distribution of the ratio from field samples above 3 μM Si(OH)₄ from the two cruises. The solid line depicts the median 3,402.

Proof of concept in culture

Cultures of Pseudo-nitzschia multiseries, Chaetoceros socialis (CCMP 172) and Chaetoceros didymus were used to test the agreement between total PDMPO incorporation and single cell incorporation measurements. The three cultures were grown the same way as the cultures for the incorporation ratio experiments: at ∼16° C in polycarbonate bottles for 24 hours on a 14/10 hour light-dark cycle. Two measurements of the average amount of PDMPO incorporated by a cell, one calculated as the quotient of the total PDMPO incorporation and cell abundance, and the other from single cell fluorescence using microscopy, show close agreement, with disparities ranging from 3 – 19% among the three species (Table 2).

Table 2.

Comparison between total community incorporation and single cell incorporation in culture. Range indicates standard deviation.

Species	Total incorporation (nmol PDMPO)	Average PDMPO incorporation from total incorporation (amol PDMPO cell-1)	Average PDMPO incorporation from microscopy (amol PDMPO cell-1)	Ratio Microscopy: Total
Pseudo-nitzschia multiseries	16.2 ± 0.27	153 ± 2.54	148 ± 131	97%
Chaetoceros socialis (CCMP 172)	10.2 ± 1.3	38.2 ± 4.8	31.0 ± 15.7	81%
Chaetoceros didymus	2.99 ± 0.32	81.6 ± 35.4	82.9 ± 25.1	99%

Proof of concept in field samples

Here we present the results from one sample from the DYEatom cruise as an example of a field application for PDMPO. This sample was collected from just below the surface with a Niskin rosette. Ambient silicic acid concentration was 5.7 μM; the incubation was performed with Si amendment to 23.8 μM [Si(OH)₄]. Initial biogenic silica concentration was 5.6 μM. Replicate samples were incubated under the same conditions with either additions of PDMPO, ³²Si, or a no-addition blank. A Bouin's preserved sample was not saved, thus to determine cell abundance and identify the diatom genera present, a subsample of the methanol preserved aliquot was counted in a 10 mL Utermöhl settling slide. Pseduo-nitzschia spp. dominated the diatom community, accounting for 82% of all diatom cells in the sample. Chaetoceros spp. made up 15% the community, while Leptocylindrus danicus accounted for 3%. Other genera were present in the assemblage but were numerically rare. Over the course of the incubation 0.68 nmol of PDMPO was incorporated, which yields a rate of silica production within 4% of that measured by ³²Si when using the culture bSiO₂:PDMPO mole ratio of 2,916.

PDMPO incorporation varied by a factor of three among diatom groups. The average cellular PDMPO content for Pseduo-nitzschia spp. in the sample was 1.9 × 10^-7 nmol cell^-1 (n=33 chains) with a standard deviation of 1.1 × 10^-7 nmol. On average, a single cell of Leptocylindrus danicus incorporated only 42% as much PDMPO, 0.8 × 10^-7 ± 0.7 × 10^-7 nmol cell^-1 (SD, n=6 chains), as a cell of Pseudo-nitzschia spp. The numerically rare, Thalassiosira spp. incorporated slightly more PDMPO than Pseudo-nitzschia spp, 2.1 × 10^-7 ± 0.4 × 10^-9 nmol cell^-1 (SD, n=2 chains). The high standard deviations relative to the mean is expected as cells were not uniformly labeled with different cells depositing different parts of the frustule.

The total community PDMPO incorporation for the field experiment was compared to the total community incorporation extrapolated from cell counts and the PDMPO incorporation by the dominate diatom genera. Total community PDMPO incorporation reconstructed from single cell measurements is 9% greater than the total community estimate. Single cell incorporation suggests that 0.75 ± 0.44 nmol of PDMPO was incorporated by the numerically dominant diatoms (i.e. Pseduo-nitzschia spp., Chaetoceros spp., Leptocylindrus danicus) over the course of the incubation, while the totally community incorporation measured 0.68 nmol PDMPO. Using the mole to mole ratio from culture, 0.75 nM of PDMPO translates to 2.2 μmol Si L^-1 of new bSiO₂, which is about 9% higher than the 2.0 μmol Si L^-1 measured with ³²Si.

Discussion

The PDMPO method is unique in that it allows for quantitative, taxon-specific silica production measurements in natural diatom assemblages. Results from the DYEatom cruise highlight some of the unique abilities of the method where the role of individual diatom genera within a mixed community can be examined using single-cell PDMPO incorporation. A genus' contribution to total PDMPO assimilation and thus total biogenic silica production, can be calculated using single cell incorporation and cell abundance data. Pseudo-nitzschia spp. accounted for 98% of the community PDMPO, which is a larger fraction of production than numerical abundance, 82%, would suggest. The Thalassiosira spp. also accounted for more of the community PDMPO, about 30% more than their numerical abundance. Conversely, Leptocylindrus danicus accounted for only 1% of the community PDMPO compared to its 3% numerical abundance. Such comparisons begin to reveal the contribution of individual taxa in a naturally diverse system informing how Si production is partitioned. This will enable experiments to examine species-specific responses to shifts in silicic acid concentration and how competition for Si may drive diatom species succession. Such understanding will inform selection of physiological parameters used in biological and biogeochemical chemical models especially those employing trait based or taxa specific approaches (Pokras and Mix 1985).

With previous PDMPO methods, community silicification could be measured quantitatively, but contribution of different diatom groups could only be inferred qualitatively. The present method overcame key analytical issues to make PDMPO fluorescence a quantitative measurement for both total community and single cell silicification. Dissolving frustules in hot NaOH (LeBlanc and Hutchins 2005) degraded the dye, so a dissolution method using low molarity HF at room temperature was devised. Accurate measurement of single-cell PDMPO fluorescence required eliminating the residual signal from photopigments by imaging with both a short and a long wavelength excitation. Converting the raw single cell fluorescence to PDMPO was accomplished using a novel fluorescence standard material consisting of uniformly labelled diatoms. The fluorescence of the standard can be quantified on both a fluorometer and on a microscope allowing RFU from microscopy to be converted to the amount PDMPO incorporated. Interspecific variation in the bSiO₂:PDMPO mole ratio was further quantified and evaluated for converting PDMPO incorporation into biogenic silica production and found to be independent of [Si(OH)₄] above 3 μM with a median of 2,916 and coefficient of variation of about 20% consistent with past findings (LeBlanc and Hutchins 2005). The ratio appears to be a linear function of [Si(OH)₄] below 3μM.

PDMPO not only provides information on silica production among taxa, but the total PDMPO incorporation measurement is relatively simple to perform at sea, allowing silica production estimates to be obtained in near real time. This method is more economical, faster and less cumbersome than attempting near-real time measurements of silica production using the recent modifications to ³²Si methodology (Krause et al. 2011). The required equipment for the PDMPO method is simple and relatively inexpensive: a fluorometer capable of the correct excitation and emission wavelengths, a refrigerator for storing samples during pigment extraction, a small centrifuge, and a drying oven. Qualitative imaging of single cells can be accomplished at sea using a variety of epi-fluorescence microscopes equipped with the appropriate filters. We chose to use a laser confocal microscope and high depth resolution to increase the accuracy of the florescence measurements.

There are a number of uncertainties with the method. Perhaps the most important is the bSiO₂:PDMPO incorporation ratio. Using cultures, we found the median ratio was 2,916 with a variance of nearly 30%. The ratio was independent of the growth rate of the cultures (R² = 0.01) ruling our growth rate as a dominant source of variation. The ratio determined from the Model II regression of natural assemblages, which includes taxa not examined in culture, falls within the confidence interval for the culture-determined ratio and has a narrower confidence interval than the culture results. Our protocols produced an incorporation ratio that was similar to what LeBlanc and Hutchins (2005) found in their culture experiments (2,800 ± 780) and in their field work (3,045 ± 230). However, our field data suggests a linear decrease in the ratio when silicic acid concentration is below 3 μM. We recommend using the ratio determined from cultures, 2,916 to convert quantitative PDMPO incorporation to biogenic silica incorporation when silicic acid concentration is above 3 μM because confounding factors like detection limit, detrital matter, radiolarians and grazing were minor or eliminated in the laboratory experiments. When silicic acid concentration is below 3 μM, we recommend using the linear relationship: bSiO₂:PDMPO = 912.6 × [Si(OH)₄] determined from the field data.

The decrease in the bSiO₂:PDMPO incorporation ratio below 3 μM may be related to shifts in diatom silicon physiology at low substrate concentrations. PDMPO may label newly deposited silica by replacing some of the organic molecules occluded within the frustule potentially polyamines as seen with the fluorescent dye, NBD-N2 and NBD-N3 (Annenkov et al. 2010). Diatoms thin their frustules in order to maintain close to maximum growth rates in low Si water (Brzezinski et al. 1990, 2011a). A decrease in the bSiO₂:PDMPO incorporation ratio could occur if the concentration of organic molecules used to organize polymerization in the SDV remains fairly constant while silica content decreases. If this were true, the changing ratio could signify frustule thinning under Si stress. Such mechanistic explanations will remain speculative until the biochemistry of silicification and the precise role of occluded organic molecules within diatom frustules are better understood.

Determining the average cellular PDMPO fluorescence for different diatom taxa is another source of uncertainty. Single cell fluorescence varied substantially among cells within the cultures imaged for the proof of concept section (see above). This does not represent measurement error, rather it is associated with the inherent variability in the Si content of the various structures being deposited by individual cells during the incubation in this largely asynchronous culture; such conditions would also be expected in a field diatom assemblages. This led to a standard deviation that was 90% of the average cellular PDMPO incorporation within the P. multiseries culture. The standard deviation in the fully labeled T. weissflogii culture was much lower, 30%. When the measurement goal is a mean incorporation rate we recommend imaging 20 cells or chains for the most abundant diatom taxa, which we find lowers the standard deviation to <100% of the mean. The number of cells that it is practical to image is dictated by the time and expense of searching for cells. As cells become rarer, it may be possible to only image 5-10 cells. Durkin et al. (2012) observed that PDMPO fluorescence of a relatively rare centric diatom (>40 μm) was three orders of magnitude higher than a small, abundant Fragilariopsis sp. within the same sample; therefore, these large rare cells could contribute substantially to total community PDMPO incorporation. Despite the high uncertainty associated with imaging fewer rare cells, we note that this variability contains useful biological information and can be exploited to examine cell synchrony, the timing and sequence of frustule development, and division rates (Shipe and Brzezinski 1999).

The PDMPO method is a fairly sensitive means of assessing total community silica production rates. The detection limit of the method was determined by the variability of the unlabeled blanks. Twice the standard deviation of the blank values corresponds to a detection limit of 0.07 μmol Si L^-1 d^-1. Thus, total community silica production can easily be measured in productive coastal waters like Monterey Bay, where silica production ranges from 1 – 30 μmol Si L^-1 d^-1 (Brzezinski et al. 1997) and the Southern Ocean where silica production ranges from 0.25 – 1 μmol Si L^-1 d^-1 (Brzezinski et al. 2001). The method is currently less adept for measuring silica production in the open ocean. Silica production rates at Bermuda Atlantic Time-Series station in the Sargasso Sea average 0.06 μmol Si L^-1 d^-1 (Brzezinski and Nelson 1995) and those at the Hawaii Ocean Time series station ALOHA average 0.03 μmol Si L^-1 d^-1 (Brzezinski et al. 2011b) which are both at, or just below the detection limit of the current method. It is likely that sensitivity will scale linearly with sample size such that, increasing sample volumes from the current 306 mL to a liter or more would allow production measurements in oligotrophic waters albeit at a higher cost per sample (i.e. three-times as much dye required to get same labeling concentration). It is important to note that the greatest power of the PDMPO method is the ability to measure production by individual cells. Oligotrophic waters do not present an analytical challenge in this regard.

The sensitivity of the PDMPO may also be increased by adding an ionophoric antibiotic rinse in an attempt to remove PDMPO from other organisms within the samples such as dinoflagellates. Despite the methanol rinse we observed PDMPO labeling in non-diatom cells in some of our samples. Alvarado (2012) found that rinsing a sample with ionophoric antibiotics reduces the fluorescence of PDMPO in dinoflagellates by 70% without significantly changing the fluorescence of the PDMPO bound in diatom frustules. Similarly, using the HF dissolution method we found rinsing with the same ionophores, monensin and nigercin (10 μM), reduced PDMPO fluorescence in a culture of Pyrocystis fusiformis incubated in f/2 media with 0.157 nM PDMPO for 24 hours to levels that were not significantly different than the blank (t-test, p=0.12, n=9). For future work, we suggest adding an ionophore rinse as an extra step in filtering the total community measurement as well in preparing the slides for single cell incorporation.

Comments and Recommendations

One of the key components of this method is the conversion of PDMPO to biogenic silica using the bSiO₂:PDMPO mole ratio. The cause of variability in this ratio within and among diatom species is not well understood. Future work might consider obtaining isolates of dominant species for culture experiments from assemblages being studied in the field to better contain the incorporation ratio specific to a study site. Or, if ³²Si is being used to measure silica production, a site specific ratio can be determined specific to each sampling location by comparing bulk PDMPO incorporation with ³²Si production.

PDMPO has the potential to be a powerful tool for understanding diatom physiology, controls on silica production, diatom resource competition and global Si cycling. Additionally, this method can be used to estimate kinetic parameters for Si production, i.e. K_s and V_max, for individual diatom taxa. PDMPO can be used to study directly the morphologic shifts such as frustule thinning (Brzezinski et al. 1990) or thickening in Fe limited water (Hutchins and Bruland 1995; Takeda 1998). Knowledge of the contribution of individual diatom taxon to overall community silica production rates will further refine production models to increase our understanding of the biological and silica pumps in the ocean.