Hydrocasts: CTD and Rosette Sample

During each cruise, a minimum of four hydrocasts were performed to collect a suite of core monthly observations. Additional hydrocasts were performed for specific process studies. We conducted separate shallow and deep casts to obtain better vertical resolution for chemical observations, and for productivity and pigment observations. Water was collected with a SeaBird rosette equipped with 12 (8 liter) teflon-coated Niskin bottles (bottle springs were also teflon-coated) at 20 depths between the surface and 1310 m. The rosette housed the CTD, which collected continuous profiles of temperature and salinity. The CTD also had a SBE-43 oxygen probe, a Wetlabs ECO fluorometer outfitted for chlorophyll-a estimates, and a C-Star transmissometer (660 nm, Wetlabs). Beam attenuation measurements were added to the time series on its 11th cruise (November 1986) originally using a SeaTech transmissometer. The rosette was controlled with a SeaBird deck unit via conducting cable, but alternatively it had been actuated automatically based on pressure recordings via an Autofire Module (SBE AFM) when breaks in cable conductivity had occurred.

Between November 1995 and September 1996, three separate SBE-19 CTDs were used in repeated casts until a reliable salinity profile was obtained below the oxycline. The SBE-19 model CTDs frequently failed to provide reliable conductivity values below the oxycline in the Cariaco Basin. Starting in September 1996, the SBE-19 CTDs were replaced by SBE-25 CTDs, which provided extremely accurate and reliable data in anoxic waters.

All CTDs were calibrated at the Sea-Bird factory once per year. The accuracy of the pressure sensor was 3.5 m and had a resolution of 0.7 m. The temperatures accuracy was 0.002°C with a resolution of 0.0003°C. The conductivity accuracy was 0.003 mmho/cm with a resolution of 0.0004 mmho/cm.

Salinity

Continuous salinity profiles were calculated from the CTD measurements. Discrete salinity samples were analyzed using a Guildline Portasal 8410 salinometer standardized with IAPSO Standard Seawater, with a precision of better than ± 0.003 and a resolution of 0.0003 mS/cm at 15° C and 35 psu, the accuracy was ±0.003 at the same set point temperature as standardization and within -2° and +4°C of ambient. These salinity values were used to check, and when necessary calibrate, the CTD salinity profiles.

Discrete Oxygen

Continuous dissolved oxygen (O2) profiles were obtained with a SBE-43 Dissolved Oxygen Sensor coupled to the SBE-25 CTD. Discrete oxygen samples were collected in duplicate using glass-stoppered bottles and analyzed by Winkler titration (Strickland and Parsons, 1972, as modified by Aminot, 1983). The analytical precision for discrete oxygen analysis was ±3 mM, based on analysis of duplicate samples, with a detection limit of 5 mM.

Nutrients

Since CAR-072 (November 2001) all samples had been filtered through a 0.8 µm Nucleopore filter within minutes of collection, as recommended by the JGOFS protocol, and frozen in plastic bottles until analysis at the University of South Florida (USF). Previous to November 2001, nutrients were filtered through a 0.7 µm GF/F filter before freezing. This data was still considered reliable, as tests using glass fiber filters show no significant contamination. The analyses follow the standard techniques described by Strickland and Parsons (1972). USF follows the recommendations of Gordon et al. (1993) for the WOCE WHP project for nutrient analysis.

Since CAR-069 (August 2001) all silica samples were kept unfrozen; they were refrigerated and kept in the dark. Prior to CAR-069, silicates were frozen and those exhibiting high concentration of silica (> 40µM below 300m in CARIACO) were affected by polymerization. All deep samples that were frozen showed low values due to polymerization loss, except CAR-063 and CAR-068 which showed high values. CAR-069 was analyzed by Yrene Astor at EDIMAR from the separate unfrozen bottlesand at USF from other, frozen, bottles. Unfrozen CAR-069 resulted higher with deep values close to what was expected (e.g. ∼92µM at 1310m).

Detection limits for CARIACO nutrient analysis

The limits below were determined by calculating the concentrations in triplicate standards, averaging the results within each triplicate group, calculating the standard deviation for each group, averaging the standard deviations, and finally doubling the averages to get the detection limits. These samples were analyzed on an ALPKEM RFA II. Subsequent Cariaco analyses were performed on a Technicon Analyzer II

Nutrient Type	ALPKEM RFA II	Technicon Analyzer II
Nutrient Type	Detection limits	Errors of analysis	Detection limits
PO₄Phosphate	0.03 µmol	<0.01 µM	0.02 µM
Si(OH)₄ Silica	0.14 µmol	0.2 µM	0.4 µM
NO₃ Nitrate	0.06 µmol	0.02 µM	0.04 µM
NO₂ Nitrite	0.02 µmol	<0.01 µM	0.01 µM
NH₄ Ammonium	0.07 µmol	0.05 µM	0.1 µM

Primary Production

Primary productivity measurements were made using a modified Steeman Nielsen (1952) NaH14CO3 uptake assay. The productivity measurements consisted of in situ incubations of water collected at 8 depths and inoculated with 14C-labeled bicarbonate. One hour before sunrise, a shallow cast was performed to obtain water from 1, 7, 15, 25, 35, 55, 75, and 100 meters. As the productivity cast was taken, a Licor Photosynthetically-Active Radiation (PAR) integrator, placed high above the ship's bridge, was activated. Water was poured directly from the Niskin bottle under low light conditions into 250 ml clear polycarbonate bottles. These bottles had been previously acid-washed, rinsed, and soaked in de-ionized water for over 48 hours. Bottles were rinsed three times before filling, a near total fill (the volume within the bottles was actually 290 ml of sea water). Four clear polycarbonate bottles were filled from each depth. We wrap one inoculated bottle from each depth in aluminum foil to obtain the dark 14-C uptake rates. An extra bottle for 1, 15, 35, and 75 m was filled, but not inoculated, to provide time-zero (t0) filter and seawater blanks. The t0 samples were kept in the dark in the laboratory and were filtered after deploying the floating incubation buoy.

We inoculated each sample under low light conditions with 1,000 ml (4 mCi) of the 14C sodium bicarbonate working solution. A 200 ml aliquot for counting total added 14C activity was removed from one of the 3 bottles from each depth and placed in a 20 ml glass scintillation vial containing 250 ml ethanolamine. The mixture was held at 5°C until subsequent liquid scintillation analysis on shore. We also placed 50 ml of the 14C working solution in a vial with ethanolamine (250 ml) for reference counting.

The dark bottle and 3 light bottles were hooked together with a combination of plastic tie wraps and nylon cord, and kept in the dark while preparations were made for deployment of the productivity incubation float. At approximately 07:00 hours, the productivity array was deployed. The entire productivity ensemble was attached to a buoy equipped with a flag and radar reflector.

Productivity observations were initiated on December 1995. Between December 1995 and November 1996, we incubated samples from 06:00 to 10:00 hours. Starting December 1996, we changed our protocol to incubate between 07:00 and 11:00 hours. This more accurately represents 1/3 of the daily photoperiod and 1/3 of the total energy received in one day on a year-round basis at 10°30'N, as verified with the PAR light sensor.

Approximately 4 hours after deployment, the productivity array was recovered. We decided to use 4-hour incubation periods due to the potentially high productivity (>1,000 mg/(m²d)) of this continental margin. Sample bottles were detached from the line and placed in labeled, dark plastic bags until filtration. Time and position of recovery were recorded. Maintaining low light conditions, a 50 ml aliquot was withdrawn from each productivity bottle using a 50 ml plastic syringe. This aliquot was filtered onto a 25 mm Whatman GF/F glass fiber filter, maintaining vacuum levels of ∼1/3 atm. The filter was rinsed with 0.25 ml 0.5 N HCl, and placed in a 20 ml glass scintillation vial, covered, and held at 5°C until subsequent processing on shore. At the shore laboratory, immediately upon return and within 15 hours of sample collection, 10 ml of liquid scintillation cocktail were added to the vials with the filters. These vials were refrigerated until they were ready for analysis on a BetaScout (PerkinElmer)scintillation counter.

Carbon uptake calculations followed the standard formulation outlined in the JGOFS manual (UNESCO, 1994), taking into consideration a (very low) quenching curve. Specifically, we subtracted the blank from all bottles, and then subtracted the dark bottle uptake from the average uptake in the light bottles to correct for non-photoautotrophic carbon fixation or absorption. Dark uptake values had always been very low. A scaling factor (∼3) was applied to convert the hourly production value to a "daily mean hourly average". This factor varies slightly, as it was based on the fraction of the energy received during the incubation period relative to the total energy received in a day. Daily rates were derived by multiplying the hourly rate by 12. Gieskes and Van Bennekom (1973), Peterson (1980), and Carpenter and Lively (1980) review the historical background, problems, and assumptions involved in the application of the radiocarbon technique to aquatic productivity. Muller-Karger (1984) also summarizes the technique and corrections involved.

pH and Alkalinity

pH samples were collected directly in 10-cm cells and analyzed on board. We measured pH and total Alkalinity estimates using the precise spectrophotometric dye methods developed by Robert-Baldo et al. (1985), Byrne and Breland (1989), and which we modified from Clayton and Byrne (1993) and Breland and Byrne (1993). These methods circumvent the problem that arises when potentiometric electrodes were transferred from dilute buffers to sea water samples due to the sample's high ionic strength. All the pH values were reported in the Master file for CARIACO data at 25 ℃ to avoid the effects of temperature on the solution chemistry. Measurement analytical precision for pHT at 25°C (total_hydrogen_ion_scale) = ±0.003, and for Total_alkalinity (mmol/kg), the precision is = 5 mmol/kg.

Corrections of pH for dye indicator impurities: The pH method uses the dye meta-cresol purple (mCP) as the pH indicator. The mCP dye used in CARIACO was in its unpurified form. Impurities in the indicator dye may cause uncertainty in measured pH values (Yao et al., 2007). Unpurified forms of the dye absorb significantly at the wavelength of maximum absorption for the acid species, HI- (434 nm) (Liu et al., 2011). The ratio of indicator absorbance at wavelengths 578 (base specie, I2-) and 434 (R = A578/A434) is used to calculate pH. Therefore, the effect of the impurities translates into apparent lower pH calculated values, especially at surface waters where pH > 8.0 (Yao et al., 2007). The effect of the impurities varies from one indicator manufacturer to another, and from different batches of the same manufacturer (Yao et al., 2007). Fortunately, the indicator used for the whole dataset in CARIACO came from the same batch. Hence, a correction for mCP impurities was applied following the method developed by Douglas and Byrne (2017) to each set of data for each cruise. This correction translated to ~ -0.01 units at pH ~ 8.1, decreasing to ~ - 0.008 units at pH ~ 7.6. The corrections were applied to the whole dataset, and values for DIC and fCO2 were recalculated in the Master file. All the pH values were reported in the Master file for CARIACO data at 25 ℃ to avoid the effects of temperature on the solution chemistry.

Chlorophyll

Chlorophyll sample collection and storage: water samples were collected from Niskin bottles into 1 L dark polyethylene bottles. These samples were immediately filtered through 25 mm Whatman GF/F filters using a vacuum of less than 100 mm Hg. During the upwelling season (approx. January-May) we filtered 250 ml seawater, and during the rest of the year we filtered 500 ml. Three replicates were taken per depth during the upwelling season, but only two were collected when biomass was obviously at its minimum, during the non-upwelling season. Filters were folded in half twice and placed in glass centrifuge tubes, labeled and frozen. Storage time was kept as short as possible (less than a week) before measurement.

Chlorophyll procedure: after removal from the freezer, the filters were extracted in 10 ml of methanol. The samples were allowed to extract for 24 hours in the refrigerator. Following extraction, samples were centrifuged for 20 minutes to remove debris. The fluorometer (Turner fluorometer model 10-AU-005) was allowed to warm up and stabilize for 30 minutes prior to use. Pure methanol was measured to confirm the zero position. Samples were transferred to 1-cm cells and they were measured directly into the fluorometer (Fo). 100 µl of 0.48N HCl was added to each cell. A second reading was taken from the fluorometer for each cell (Fa). Standardization. The fluorometer was calibrated every year with a commercially available chlorophyll a standard (Σ). The concentration of chlorophyll-an and phaeopigments in the sample were calculated using Yentsh and Menzel (1963) equation, with a specific absorption coefficient of 74.5 (chlorophyll in methanol).

HPLC

HPLC analysis was restarted in July 2006 (CAR-123). Samples were filtered 47 mm Whatman GF/F filters at 8 depths (1, 7, 15, 25, 35, 55, 75 and 100m). The volume filtered depends on the amount of particles in the water. Replicates were taken at the 1m depth. Filters were stored in aluminum envelopes and stored in the fridge until reaching shore. Once on shore, samples were stored at -40°C until transportation to the US. Horn Point Laboratory (http://www.hpl.umces.edu/) performs the analyses through a collaborative agreement with NASA. The method used was Van Heukelem and Thomas (2001).

POC and PON

POC and PON sample collection and storage: water samples were collected from Niskin bottles into 2 L dark polyethylene bottles. These samples were immediately filtered through 25 mm Whatman GF/F filters (precombusted for 5 hours at 450°C) using a vacuum of less than 100 mm Hg. Since July 2007 (CAR-135) filters were acidified (10% HCl) after combustion and prior to sample collection. A portion of these filters was used for POP analysis (see below). Filters were placed on expendable tin disks and then into aluminum foil envelops (also precombusted for 5 hours at 450°C) labeled and frozen. In the laboratory, filters were dried at 65°C for 12-15 hours then stored with silica gel.

Measurement: The filters were folded inside a tin disk and analyzed on a Perkin Elmer 2400 Elemental Analyzer. The samples were combusted at 1200-1300°C and then passed through a reduction tube to removes the oxygen added to raise the combustion temperature. Filers were not acid fumed prior to analysis. The C and N were then separated in a chromatographic column and were measured on a Thermal Conductivity Detector. Carbon and nitrogen standards, and blank filters were used to calibrate the data. The accuracy of the instrument was <0.3% and the precision of the instrument was <0.2%. These were published values and we find that we were always within these limits (usually ±0.15% for carbon and ±0.1% for nitrogen). We ran cystine as our standard (29.99% Carbon, 11.66% Nitrogen). The analytical range of the instrument is: Carbon= .001 to 3.6 mg and Nitrogen= 0.001 to 6.0 mg.

POP

POP was analyzed from the same POC/PON filters. The method used for the SRP analysis was based on Koroleff (1983).

Dissolved organic Carbon, Nitrogen and Phosphorous (DOC, DON and DOP)

Measurements of DOC were taken since the beginning of the project in November 1995 but suspended in February 2001 (CAR-062) due to irregularity of results. DOC was reinitiated in March 2005 (CAR-110) using a new protocol. DOC samples were collected monthly and analyzed at the Organic Biogeochemistry Lab in the Rosenstiel School of Marine & Atmospheric Science at the University of Miami. Samples were gravity-filtered directly from the Niskin bottles through 45 mm GF/F precombusted filters using acid cleaned polycarbonate in-line filter holder. Immediately after filtration the polyethylene bottles were frozen at -20°C until analysis.

DON and DOP measurements were added to the regular CARIACO cruises in July 2004 (CAR-101). Samples were filtered through GF/F filters using a specially built vacuum filter rack. The DON method was based on Solorzano and Sharp (1980). This procedure produces a filtered seawater sample for analysis of total dissolved fixed nitrogen (=nitrate + nitrite + ammonium + DON). DON concentration was obtained by difference from nitrate, nitrite, and ammonium measured in the standard nutrient protocol. DOP was analyzed in the same persulfate-oxidized filtrate solution as DON. That solution yields total inorganic phosphate concentration, which was composed of the inorganic phosphate concentration originally in the seawater, plus an additional phosphate concentration due to the conversion of DOP to phosphate. DOP concentration was then obtained by difference from the inorganic phosphate in the unoxidized sample measured through the standard nutrient protocol.

Optical measurements

In-water measurements: a PRR-600 (Biospherical) was used to retrieve downwelling irradiance and upwelling radiance. From these, PAR, Kd and reflectance can be calculated. Beam attenuation coefficient (Cp) was measured using a C-Star transmissometer (see section Hydrocasts: CTD and Rosette Sample), which provides measurements at 660 nm throughout the entire water column.

CDOM samples were measured at four depths (1m, 15m, 25m, 50m) by filtration through a 0.2 µm pore size filter and immediately frozen at -20°C. Before being analyzed, they were thawed and re-filtered to eliminate any salt crystals that may had formed. CDOM was measured between 200 and 800 nm, with a 0.3 nm interval, using a dual fiber optic spectrometer (Ocean Optics) equipped with 10-cm quartz cuvettes and distilled water as a blank.

Above water measurements: a PR-650 (Photoresearch) measures sky radiance (Ls), water leaving radiance (Lw) and total irradiance (Es) at an angle of 30°. From these measurements, remote sensing reflectance (Rrs = Lw/Es) can be calculated and used in satellite sensor (such as MODIS and SeaWiFS) calibration.

Methods compiled by John Akl, July 2002. Revised November 2005 by Laura Lorenzoni. Revised April 2019 by Digna Rueda-Roa

The CARIACO Ocean Time-Series Program (November 1995 – January 2017)

For a detailed log for each cruise, please refer to the supplemental document Cruise Data Acquisition Report (https://datadocs.bco-dmo.org/docs/302/CARIACO/data_docs/3092/1/Cruise_data_aquisition_report.xlsx)