Methodologies as extracted from the
Subsets of CTD data taken at the time of water sample collection (a 10 second average) were transmitted to the bottle data files immediately after each cast to provide pressure and temperature at the sampling depth, and to facilitate the examination and quality control of the bottle data as the laboratory analyses were completed. The discrete hydrographic data were entered into the shipboard data system and processed as the analyses were completed. The bottle data were brought to a usable, though perhaps not final, state at sea.
Salinity samples were drawn into ODF salinity bottles which were rinsed three times before filling. Tests have shown that these sample bottles are capable of routine high quality performance over storage times of several days. Salinity was determined after sample equilibration to laboratory temperature, usually within about 8-36 hours of collection, by conductivity ratio analyses with a Guildline Autosal 8400 laboratory salinometer. A minimum of two values were recorded for each sample, after flushing the Autosal cell several times before the first reading and then again between readings. The salinometer was standardized against a fresh vial of Wormley/IAPSO Standard Seawater, batch P108, at the beginning and end of each box of samples, usually one cast, and extra vials were opened in any case where the results were suspicious.
Salinity was calculated according to the equations of the Practical Salinity Scale of 1978 (UNESCO, 1981) from the conductivity ratio and Autosal bath temperature. Accuracy estimates of bottle salinities run at sea are usually better than 0.002 psu relative to the specified batch of standard. Although laboratory precision of the Autosal can be as small as 0.0002 psu when running replicate samples under ideal conditions, at sea the expected precision is about 0.001 psu under normal conditions, with a stable lab temperature. On this cruise, the salinometer was installed in a small interior lab space, with a reasonably stable ambient temperature.
Dissolved oxygen samples were titrated in the volume-calibrated iodine flasks with a 1 ml microburet, using the whole-bottle Winkler titration following the technique of Carpenter (1965). Standardizations were performed with 0.01N potassium iodate solutions prepared from preweighed potassium iodate crystals. Standards were run at the beginning of each session of analyses, which typically included from 1 to 3 casts. Several standards were made up and compared to assure that the results were reproducible, and to preclude basing the entire cruise on one standard, with the possibility of a weighing error. A correction (-0.014 ml/l) was made for the amount of oxygen added with the reagents. Combined reagent/seawater blanks were determined to account for oxidizing or reducing materials in the reagents, and for a nominal level of natural iodate (Brewer and Wong, 1974) or other oxidizers/reducers in the seawater.
The quality of the KIO3 is the ultimate limitation on the accuracy of this methodology. The assay of the finest quality KIO3 available to ODF is 100%, ± 0.05%. The true limit in the quality of the bottle oxygen data probably lies in the practical limitations of the present sampling and analytical methodology, from the time the rosette bottle is closed through the calculation of oxygen concentration from titration data. Overall precision within a group of samples has been determined from replicates on numerous occasions, and for the system as employed on this expedition, one may expect ± 0.1 to 0.2%. The overall accuracy of the data is estimated to be ± 0.5%.
Phosphate was analyzed using the hydrazine reduction of phosphomolybdic acid as described by Bernhardt & Wilhelms (1967). Silicate was analyzed using stannous chloride reduction of silicomolybdic acid. Nitrite was determined using diazotization and coupling to form dye; nitrate was reduced by copperized cadmium and then analyzed as nitrite. These three analyses are described in Armstrong et al., (1967). Sets of 4-6 different concentrations of shipboard standards were analyzed periodically to determine the linearity of colorimeter response and the resulting correction factors.
Interpreting the precision of nutrient analyses must be done with some care. The precision of identical samples run consecutively in the same AutoAnalyzer run will typically be better than the same samples run in different batches. Nutrient analyses, particularly silicate, are temperature sensitive, and for the best results require a constant temperature laboratory. During cruise 119.4 (Reykjavik to Azores), the AutoAnalyzer was installed in a laboratory on the first platform deck, with no outside access, and with a stable air temperature. During the next leg, the AutoAnalyzer had to be moved to the main lab, with traffic directly to the outside producing greater variability in air temperature. The overall precision is estimated to be ± 0.5% of the maximum value for a given cast, with a cast-to-cast precision of ± 1-2%, and an accuracy of 2-3%, with somewhat better results expected for cruise 119.4.
Thymidine incorporation samples collected from Niskin rosette casts were immediately processed as described in Ducklow and Hill (1985), with the following modifications: Samples were incubated with 5 nM 3H-thymidine (New England Nuclear, sp. act. 81 Ci/mmol) in polycarbonate bottles, disposable polyproplyene centrifuge tubes or Whirl-Pak bags. Incubations were terminated with addition of 0.37% formaldehyde, then filtered onto 0.2 µm Nuclepore filters. Extractions were carried out by rinsing each filter on its funnel support 3 times with 5% ice cold TCA, over a weak vacuum (<10 in. Hg), then 3 times with 80% ice cold ethanol. All extracted filters were stored dry in scintillation vials for counting at Horn Point Environmental Laboratory.
Leucine incorporation samples were treated according to the method described in Kirchman et al., (1985), with the following modifications: Samples were incubated with 0.5 nM 3H- leucine (NEN; Sp. Act. 73 Ci/mmol) and 10 nM nonradioactive leucine, then treated as described for thymidine.
For shipboard analysis of suspended particulate matter samples, we use a 10 cm reverse-phase ODS column (adsorbosphere HS, 3µm) eluted with a linear gradient of 0-100% B (A=20/80 0.5N NH4Ac(aq)/MeOH; B=20/80 acetone/MeOH) over 10 min at 1.5 mL/min. For more detailed analysis, we use a 15 cm column and a 30 min gradient. All samples were analyzed by diode array spectroscopy (350-700nm) and fluorescence detection.
Individual components were quantified by normaling the data to the internal standard to correct for evaporative losses, differences in water retention by filters, and nonquantitative transfers during handling, then correcting the data for individual compounds by using an appropriate detector calibration factor. We calibrate our system by isolating pure pigments, making quantitative standards, and carrying out a series of analyses over the calibration range of interest. Depending on the pigment's spectral properties, and the wavelength of detection, calibration factors for individual pigments may differ by up to a factor of 5. For calibration in the field, concentrated solutions of standards are prepared in benzene, and 1 mL aliquots dispensed to 10 mL volumetric flasks. The flasks are flushed with N2, capped, frozen at -20° Centigrade. Randomly chosen standards are stored in the laboratory for spectroscopic analysis at a later date. The remaining flasks are stored on shipboard until use. For calibration, the standards are returned to room temperature and acetone or methanol added to 10 mL. Aliquots are then analyzed by HPLC. In the US JGOFS program, over 30 replicate standards were prepared and analyzed daily in triplicate showed a SD of <10% by HPLC analyses. Detectors were calibrated daily for the internal standard and twice weekly for chlorophyll-a. Calibration factors for major xanthophylls were determined at least one time per week.
In order to calculate the partial pressure of CO2 at temperatures other than 20.00° C, a temperature coefficient of 4.23%/°C should be used (Chipman et al., in prep.). The following equation can be applied:
pCO2(T) = pCO2(20) X exp(.0423 * (T - 20))Total CO2 Concentration-
Samples for TCO2 measurements are drawn from the Niskin bottls of the rosette casts into 500 ml Pyrex bottles with ground glass standard-taper stoppers, sealed with silicone vacuum grease. All samples were poisoned with 250 µl of 50% saturated mercuric chloride solution to prevent biological alteration of the sample and were analyzed within 48 hours of collection.
For analysis, the seawater was introduced into a fixed- volume pipet (approximately 25 ml) by means of a gravity feed and at least 5 volumes were allowed to flush the pipet before the final volume was isolated and transferred to the extraction chamber. The sample was acidified with 1/2 ml of 8.5% phosphoric acid while in the pipet, and after draining the pipet was swept by the carrier gas for one minute to assure all of the CO2 from the sample was swept into the coulometer cell for analysis. The volume of the sample pipet had been previously determined by weighing empty and filled with deionized, distilled water, using the weight difference (corrected for the buoyancy of air displaced by the water) and the density of pure water at the temperature of filling to give the volume of the pipet (on a "to contain" basis).
The calibration factor of the coulometer (UIC/Coulometrics Model 5011) was determined several times a day, using pure gaseous CO2 injected into the carrier gas stream by means of either of two calibrated sample loops mounted on an 8-port sampling valve. The pressure of the CO2 gas was determined by venting the loops to the atmosphere and measuring the absolute barometric pressure by means of a high-accuracy electronic barometer, while the loop temperatures were measured to ±0.05 °C using a thermometer whose calibration is traceable to the N.B.S. The non-ideality of CO2 at the temperature and pressure of calibration was incorporated in the computation of the loop contents. A linear drift of the instrument calibration, which generally amounted to less than 0.2%, was applied to all seawater analyses, and blanks for the CO2 contained in the acid and the system blank were subtracted before computing the concentration of TCO2 in the samples. The sample concentrations were computed on a "per kilogram" basis using the salinity of the sample, the temperature of the sample in the pipet at the time the sample was isolated, and the International Equation of State of Seawater (Millero et al., 1980) to determine the density, after correcting for the dilution of the sample by the CO2-free mercuric chloride poison which had been added.
Total Alkalinity-
The total alkalinity of seawater was computed (NOT
MEASURED) using the following measured quantities in each sample:
the total CO2 concentration, the partial pressure of CO2 at 20° C, the salinity and the concentrations of silica and
phosphate. The total concentration of borate (TB) has been
assumed to be proportional to salinity [TB (µM/kg)= 410.6 X
(salinity/35)]. For our computation, the total alkalinity (TALK)
in seawater is defined by:
TALK = Ac + Ab + Asi + Ap + Aw, where Ac = Carbonate alkalinity = [HCO3-] + 2[CO3=], Ab = Borate alkalinity = [H2BO3-], Asi = Silicate alkalinity = [H3SiO4-], Ap = Phosphate alkalinity = [H2PO4-] + 2[HPO4=] + 3[PO4-3], Aw = Water alkalinity = [OH-] - [H+].The following equilibrium constants have been used for the computation: the solubility of CO2 in seawater measured by Murray and Riley (1971) and formulated by Weiss (1974); the first and second apparent dissociation constants for carbonic acid measured by Mehrbach et al. (1973); the first apparent dissociation constant for boric acid in seawater measured by Lyman (1956); the first apparent dissociation constant for silicic acid in NaCl solutions measured by Ingri (1959); the first, second and third apparent dissociation constants for phosphoric acid in seawater measured by Kester and Pytkowicz (1967); and the apparent dissociation constant of water and the activity coefficient of hydrogen ion in seawater formulated by Millero (1979) on the basis of determinations by Culberson and Pytkowicz (1973). The equilibrium reaction equations, computational procedures and references are given in detail by Peng et al., (1987).
A hydrocast generally to 200m was collected using a CTD rosette mounted with 30 liter Niskin bottles. Sample depths were similar on all dates, and were chosen to intensively sample the euphotic zone and representively sample beneath that region; the 200m depth was replaced by 1000m on May 22. The cast on May 31 was altered to sample at the 300, 500 and 1000 meter horizons. Collection time was 1030-1130 local time. Samples were drawn directly into replicate 1 liter polyethylene bottles and processed immediately. Two approaches were used: (1) slides were prepared for traditional visual and automated image analyzed epifluorescence microscopy of pico- and nanoplankton; and (2) samples were preserved for subsequent microscopic determination of larger, rarer microzooplankton.
Duplicate 15-25ml samples were preserved with 0.3% gluteraldehyde and stained sequentially with 5 µg/ml DAPI and 4.5 µg/ml proflavin (final concentrations). Samples were drawn under low vacuum (<5mm Hg) onto 0.2µm black Nuclepore filters, with cellulose nitrate backing filters to enhance even cell distribution. Filters were mounted on glass slides with a drop of immersion oil between the filter and glass coverslip. They were either processed immediately or frozen for subsequent analysis. Two microscope systems were set up in the gravity lab of the Atlantis II. This area was supplied with regulated power and was darkened for fluorescence microscopy. An Olympus BH-2 epifluorescence microscope with a 100 watt Hg burner was used to identify and enumerate ciliates, heterotrophic dinoflagellates, and other plankton of specific taxonomic interest. Samples were counted at either 500x (dry) or 750x (oil) using blue (proflavin) or UV (DAPI) excitation. Generally, all the cells contained on one half of each filter were enumerated. In the case of abundant organisms such as heterotrophic dinoflagellates, cells in10 paths were enumerated or 200 cells counted, whichever came first.
Samples were also analyzed using a state-of-the-art color image analyzed fluorescence microscope (CIAFM) system. For cyanobacteria, random fields were chosen under green excitation. After frame-grabbing a background image was subtracted from the red image and cells were counted and measured in this image using a global threshold, minimum and maximum size boundaries, and interactive editing. At least 7 fields and 200 cells were counted and measured per sample. For the rarer nanoplankton, transects of the slide were made under violet excitation using the computer-controlled microscope stage. After frame-grabbing, individual cells were selected and 2-color (red and green) single cell subimages were stored after background subtraction and interactive operator classification as photo- or heterotroph. Cells were automatically counted and the area of the filter examined was continuously calculated so cells per ml in the original sample could be determined. At least 150 cells per sample were measured. Analysis of the subimages was done automatically by choosing a random subset of the images for threshold determination by the MinD2 method. The mean threshold was then applied to the whole set of heterotrophic or phototrophic cells from each sample. The average of the red and green subimages were used for cell size measurements. Samples from three profiles at 47N and one at 59N were prepared for sequential epifluorescence/electron microscopy (SEEM). This method allows cells to be identified according to fluorescence characteristics and then shadow-cast for taxonomic identification using either scanning or transmission electron microscopy. About 50mls were concentrated in a swinnex filter holder, rinsed with 15-20 ml distilled water and stained with DAPI (as above). Droplets of the concentrate were micropipetted onto No. 200 mesh copper (Pinpointer) grids and dried in a vacuum desiccator. Within 36h grids were mapped under epifluorescence microscopy for the positions of hetero- and phototropic cells. One liter samples from each depth on every profile were preserved in a final concentration of 2% hexamethylamine-buffered formaldehyde for subsequent enumeration of larger, rarer microplankton which were not adequately sampled in the small volumes filtered for epifluorescent examination. On three profiles at 47N and one at 59N, triplicate one liter samples from each depth were preserved in 2% buffered formaldehyde, 10% acid Lugol's, and 5% Bouin's fixative to investigate variability in preservation and shrinkage of microplankton.
µgC/l=biomass, calculated as µm3/*F, where F is a biovolume to carbon conversion factor. We used F=220 fgC/µm3
Although fixation in 2% formaldehyde allows for epifluorescence microscopy and does not severely distort cell shape in ciliates, there are significant losses of cells during fixation (Gifford, personal comm.). To correct for these losses and to allow for identification of the more common ciliate species using protargol stain (Montagnes and Lynn, 1987), we preserved samples from three profiles in 10% acid Lugol's and in 5% Boiun's solution. Based upon comparison between 16 pairs of formaldehyde and Lugol's fixation samples, we empirically determined a correction factor for the counts of non-loricate ciliates done on the formaldehyde-fixed samples. No correction factor was applied to the counts of tintinnids and heterotrophic dinoflagellates.
Based on the corrected cell densities and average cell volumes, biomass (µg C/l) was calculated using appropriate conversion factors for ciliates (Putt and Stoecker 1989) and 0.14 fg C/µm--3 for heterotrophic dinoflagellates (Lessard, in press).