What an ocean nutrient result actually represents
An ocean nutrient number is a chemical measurement assigned to a particular sample of seawater, not a direct measurement of marine productivity or ecosystem health. Nitrate, nitrite, phosphate and silicate are analytical quantities with different operational definitions. A value becomes useful only when the reader knows the analyte, calibration, collection depth, handling history, units and quality flag. The phrase ‘nutrient-rich water’ is too vague to support quantitative conclusions without those details.
Autoanalysers often determine dissolved nutrient concentrations by controlled chemical colour reactions. Reagent and sample travel through an instrument, form a coloured product, and produce an optical signal that is compared with standards. The detector therefore observes light, while the chemist infers concentration through a tested method. A clean number in a published spreadsheet does not guarantee that blanks, standard drift, reduction efficiency or sample-storage effects were negligible.
This manual translates the analytical chain into readable decisions. It explains nitrate and nitrite subtraction, phosphate and silicate reactions, calibration slopes, matrix effects, detection limits, dilution, density conversion, profiles and uncertainty. The numerical values in the worked examples are explicitly constructed for teaching; none claims to come from a real ship expedition. Readers should be able to rebuild each calculation and refuse interpretations for which necessary metadata are absent.
Define what was dissolved and what was total
Marine nutrient chemistry includes more than the three convenient labels on a graph. Nitrogen may appear as nitrate, nitrite, ammonium, dissolved organic compounds or particulate material. Phosphorus may be inorganic or organic, dissolved or particulate. Silicon can be dissolved as reactive silicic acid species or associated with particles and biogenic silica. A standard colorimetric channel measures its own defined analyte fraction rather than all these pools at once.
Filtering changes the operational fraction. A sample passed through a specified filter soon after collection may differ from an unfiltered bottle analysed after delay. Filters can introduce contamination or adsorption; keeping the bottle open can permit changes. The report should state filtration, timing and storage conditions. A dissolved reactive phosphate value must not silently become a total phosphorus value in a translated summary, even when both are expressed using phosphorus-related units.
A high measured nutrient concentration is not automatically a high nutrient supply rate to plants. Concentration is amount per mass or volume at an instant; supply involves transport, mixing and regeneration over time. Biological availability can also depend on chemical form. A thoughtful interpretation retains the exact measured species and separates what was in the water from how rapidly it was delivered, consumed or returned.
How automated colorimetry converts a reaction into data
Continuous-flow, segmented-flow and discrete systems can all perform nutrient colorimetry, but their fluid-handling designs differ. In a segmented system, small air gaps may separate liquid portions and reduce dispersion. Reagent addition, residence time and reaction temperature affect how much coloured product forms. A photometric cell measures absorbance at the method’s wavelength, then software converts the signal to concentration with calibration data.
Beer–Lambert reasoning supplies a useful model: within a suitable range, absorbance can rise approximately linearly with the concentration of a coloured reaction product. In real autoanalyser data, colour-development yield, matrix effects and detector background also matter. An empirical calibration curve is therefore more defensible than treating the raw absorbance as a universal concentration. Check path length, wavelength, standards and curve residuals before trusting an extrapolated result.
The NOAA cruise nutrient descriptions give actual examples of a multichannel analyser measuring nitrate, nitrite, phosphate and silicate. Their reagent and wavelength details are primary method evidence for that particular workflow, not universal requirements for every modern instrument. Readers should preserve instrument identity and laboratory protocol instead of replacing them with a generic claim that ‘a spectrometer measured the nutrients.’
Nitrate-plus-nitrite is not always nitrate alone
Many analytical methods convert nitrate to nitrite first, then measure the coloured reaction associated with nitrite. Depending on the configuration, the combined channel can represent nitrate plus naturally present nitrite, often written NOx. A separate nitrite reading permits nitrate estimation by subtraction when both are on matching concentration scales. This is a difference between measurements with their own uncertainties; it is not a free correction independent of calibration.
For a constructed example, suppose NOx is 12.30 micromoles per litre and nitrite is 0.80 micromoles per litre. The calculated nitrate concentration is 11.50 micromoles per litre. If the laboratory’s delivered nitrate column already incorporates that subtraction, subtracting a nitrite column again is a processing mistake. Read the data dictionary and method notes rather than guessing from the short variable name on an axis.
Reduction efficiency matters because nitrate might not be completely converted under faulty reagent or column conditions. A precise optical reading of incompletely reacted nitrate is not a precise nitrate concentration. The laboratory can use reference standards and appropriate efficiency checks to establish performance. Anomalous negative nitrate after subtraction should be treated as a possible near-limit or processing issue, not evidence that seawater contains a negative quantity.
Reactive phosphate and what its colour does not include
Reactive phosphate methods commonly form a phosphomolybdate species that is converted into an optically detectable blue product through controlled chemical steps. The reported concentration is tied to the method’s reaction conditions and standard solutions. It is generally not a total phosphorus analysis. Low-phosphate surface waters demand special attention to bottles, rinse water and reagents because a small contaminating amount can rival the environmental concentration.
A phosphate value just above blank may be analytically uncertain even if the software displays several decimals. Reagent blank, matrix blank and detection limit are different controls. A low or negative blank-corrected estimate should not be reclassified as physically zero without an explicit reporting convention. In comparative studies, changing analytical detection capability can cause an apparent environmental shift unless censoring and method history are addressed.
Even reliably low phosphate does not prove a particular species is phosphorus limited. Organisms can access other forms, water supplies nutrients through transport, and growth can be constrained by light, temperature or other resources. A carefully worded conclusion says the measured reactive phosphate pool was low under specified conditions. Experimental or multi-variable evidence is needed before turning that observation into a causal diagnosis.
Silicate colour and chemical interference
Dissolved silicate can be quantified through a molybdate-based reaction that produces a coloured complex following specified reduction steps. The laboratory chooses reagents and reaction conditions that manage interfering responses from other substances, including related phosphate chemistry. If a reagent is omitted, old or mixed incorrectly, colour may not be specific to silicate. The detector cannot diagnose that chemical mistake by itself.
NOAA’s historical hydrographic method description discusses silicomolybdic acid, a blue reduced compound and a reagent used to impede phosphate-related colour. This illustrates why method protocols deserve to be read as more than equipment shopping lists. Different validated analytical procedures may use different reagents, optical wavelengths and concentration ranges. Do not move a calibration from one procedure to another without appropriate evidence.
Silicate is ecologically relevant to diatoms and to the movement of water masses, yet its concentration is a standing pool, not a rate of diatom growth. High values can reflect supply or remineralisation, low values can reflect uptake or a water mass’s history. A good profile interpretation includes salinity, depth, season and companion measurements rather than using one silicate maximum as a unique process fingerprint.
Calibration curves and intercepts
Suppose a photometric calibration in a constructed exercise follows absorbance A = 0.040 + 0.120 C, where C is micromoles per litre. A sample with A = 0.280 gives C = (0.280−0.040)/0.120 = 2.00 micromoles per litre. Subtracting the intercept matters because part of the observed absorbance does not arise from analyte concentration. The numerical result is meaningful only inside the validated concentration and matrix range.
A single high R-squared is not enough to establish calibration quality. Look at each standard’s residual, whether low-concentration standards were included, and whether a high standard saturates the response. Reagent strength, optical cell cleanliness, tubing changes and reaction timing can shift slope or intercept. Independent check standards at useful concentrations help determine whether the curve still represents the instrument throughout the run.
A standard prepared in pure water might not respond identically to one prepared in a seawater-like matrix. Salt composition, background colour and reaction chemistry can change the signal. Laboratories may use low-nutrient seawater or matched matrices to reduce bias. When comparing estuarine samples across a salinity gradient, a matrix assumption valid in oceanic water might need extra testing. Record the standards’ matrix rather than pretending calibration is context-free.
Blanks, controls and carryover
A reagent blank tests background from a defined portion of the chemical system. A matrix blank can include response from salt water without intentional analyte, while a procedural blank can follow more of the handling path. These are related but not interchangeable. Calling every correction simply ‘blank subtracted’ hides what was actually removed, whether the correction had concentration units or absorbance units, and whether it varied during the run.
A nutrient-rich deep sample immediately followed by a nearly nutrient-free surface sample tests the system’s washout. Tubing and valves can retain material; the subsequent peak may be biased by carryover. Bracketing blanks, wash standards or appropriate repeated samples can identify this effect. A reader should check sequence order before interpreting an isolated high value as a natural feature.
Instrument drift may produce a gradual change in reference or check-standard response while the ship passes through a geographic gradient. Plot controls against analysis time to separate the instrument’s chronological behaviour from the ocean’s spatial pattern. Corrections should be based on validated laboratory rules, not a desire to make a cruise section look smooth. If checks fail and no supported repair exists, quality flags communicate the limitation.
Detection and quantification are not identical
A detection limit concerns whether an analytical signal can be distinguished from an appropriate blank distribution, while a quantification threshold concerns whether a reliable numeric result can be assigned. Their actual values depend on method validation and sometimes matrix or run conditions. A value near the limit should not acquire greater credibility merely because a display prints three decimal places.
Imagine an illustrative phosphate quantification limit of 0.05 micromoles per litre and a computed result of 0.03. The method may allow the signal to be retained as estimated but not fully quantitative, or coded below a reporting threshold. It is not scientifically equivalent to 0.000 phosphate. The correct language is about the method’s ability to resolve concentration, not about absolute chemical absence.
Replacing all below-limit estimates with zero biases averages and ratios downward. Replacing them with half a reporting limit is a statistical modelling decision requiring explanation, not an additional laboratory measurement. If the detection threshold changes between surveys, a naive long-term trend may reflect that change. Retain censoring flags, reported thresholds and raw values where available before computing environmental statistics.
Sampling, delay and dilution
After a Niskin bottle is opened, biological uptake, remineralisation and contamination can alter nutrients. Laboratories use procedures for prompt analysis or validated storage, sometimes involving filtration, refrigeration or freezing depending on the analyte. A handling method suitable for nitrate may not transfer automatically to silicate or phosphate. The dataset should preserve collection-to-analysis time and storage treatment rather than making an undocumented assumption of perfect stability.
High-concentration samples may be diluted into the calibration range. A one-in-five dilution means one part sample in five parts final mixture and requires multiplying the diluted concentration by five when software has not already applied the factor. Saying ‘one to five’ without specifying final versus added volume creates ambiguity. Explicit measured volumes, diluent composition and processing state are safer than shorthand.
Consider a diluted aliquot reporting 8.0 micromoles per litre and a validated dilution factor of five. The original concentration is 40.0 micromoles per litre. If the final archive already lists 40.0, multiplying again fabricates 200. A quality check should compare raw signal, diluted result, dilution factor and final output. Errors in data reduction are as real as errors in wet chemistry.
Amount per litre, amount per kilogram and elemental mass
Two nutrient numbers cannot be compared without examining their denominators. A value in micromoles per litre is a volume-specific concentration; a value in micromoles per kilogram is a mass-specific concentration. Seawater density links them. If density is 1.025 kilograms per litre, 12.30 micromoles/L corresponds to 12.30/1.025 = 12.00 micromoles/kg. This simple example shows why assuming that water always weighs exactly one kilogram per litre can create a false difference between high-quality surveys.
Nutrient archives may alternatively report milligrams of nitrogen per litre, milligrams of nitrate ion per litre, or micromoles of nitrogen species. These are not interchangeable labels. A mole of nitrate ion contains one mole of nitrogen atoms, but the molar mass of the nitrate ion is much greater than that of nitrogen alone. Preserve the full chemical label such as NO3-N or PO4-P and show every molar-mass conversion in the methods appendix.
Consider a sample of 14 micromoles of nitrate ion per litre. That also contains 14 micromoles of nitrogen atoms per litre, corresponding to about 0.196 milligrams N/L using nitrogen’s molar mass of roughly 14.01 g/mol. Its mass expressed as nitrate ions would be larger. A translation that omits the N label silently changes a regulatory or ecological quantity even when the number remains on the same line.
The value and the weakness of nutrient ratios
The approximate 16:1 nitrogen-to-phosphorus ratio familiar from classic marine biogeochemistry is a reference for broad organic-matter stoichiometry, not a universal threshold proving whether a location is nutrient limited. The ratio of dissolved nitrate concentration to dissolved phosphate concentration is not necessarily the same as organism composition, nutrient supply ratio or uptake rate. All chemical forms, physical transport and ecological context matter before a limitation claim can be made.
For an illustrative seawater sample with nitrate 4.0 and phosphate 0.25 micromoles/L, the molar concentration ratio is 4.0/0.25 = 16. A second sample with nitrate 0.16 and phosphate 0.010 has the same ratio but much smaller concentrations. Those waters could differ dramatically in absolute nutrient supply and biological response. The ratio should never replace the individual concentrations, their uncertainties or the relevant organism’s needs.
When a denominator is close to detection limits, a ratio is particularly fragile. With phosphate estimated at 0.020 ± 0.010 micromoles/L, a nitrate-to-phosphate quotient can swing markedly across plausible phosphate values. This is not a trivial rounding issue. A report that declares phosphorus limitation from a spectacular ratio may have built its headline on a barely resolved denominator. Present the individual values, their censoring status and uncertainty before the quotient.
Depth profiles, water masses and sampling geometry
An ocean nutrient depth profile is a collection of bottle or sensor observations placed along pressure or depth, not necessarily a continuous chemical trace. Smooth connecting lines can create the illusion of resolution between widely separated samples. Match each concentration to the correct station, bottle, sample time and depth before interpolating. A sharp jump may be a real water-mass boundary, but an incorrectly joined sample identifier can produce a similar graphical effect.
Surface nutrient depletion can reflect biological uptake and the history of physical supply, while deeper maxima may reflect remineralisation or circulation. These explanations are plausible, not uniquely established by one concentration profile. Temperature, salinity, dissolved oxygen, irradiance, mixed-layer structure and sampling season help distinguish alternatives. One should also consider freshwater input and horizontal advection in coastal waters rather than treating all concentration gradients as vertically generated biology.
The NOAA World Ocean Atlas provides quality-controlled nutrient climatologies built from many profiles. These reference patterns help establish regional context, but a climatological average is not a measurement of what happened at an individual station on a given day. Compare location, depth, season, time coverage and gridding assumptions. A measured departure from climatology can be a real transient event, a sampling issue or ordinary variability; the climatology alone cannot decide which.
Shipboard chemistry and optical profiling floats
Biogeochemical Argo floats can carry ultraviolet optical nitrate sensors and other biogeochemical instruments. Those sensors do not use exactly the same procedure as a shipboard segmented-flow reagent analyser. UV absorption, spectral fitting and sensor drift present distinct challenges from reagents, reduction columns and colorimetric blanks. A near-match between their nitrate estimates is useful validation evidence only when water-mass, pressure and timing differences are treated explicitly.
The Argo program distinguishes rapid data delivery from subsequent scientific delayed-mode quality control. A provisional sensor profile may help scientists see an evolving feature, while subtle climate or biogeochemical trend analysis benefits from carefully adjusted records. Check whether the data field is raw or adjusted and which quality flags apply. Do not treat the availability of a decimal in real time as proof that later calibration work is unnecessary.
If a float and ship sample disagree, start with a matched-depth and matched-time comparison rather than selecting the larger value as truth. Inspect salinity, optical interferences, laboratory reference checks, instrument drift and data-processing revisions. A difference may reflect genuine spatial variability or temporal change between measurements. The scientifically defensible result is a comparison with a discrepancy budget, not an unqualified statement that one technology is inferior.
Uncertainty follows subtraction and ratio arithmetic
When nitrate is calculated as a combined NOx measurement minus a separate nitrite measurement, the resulting error depends on both measurements. Under a simplified independent-error model with uncertainties 0.20 and 0.10 micromoles/L, the combined standard uncertainty is sqrt(0.20² + 0.10²), about 0.224 micromoles/L. If the channels share calibration or blank errors, correlation may change that result. The independence assumption should be an explicit teaching simplification, not a universal rule.
An uncertainty budget can include standard preparation, reagent purity, matrix differences, detector noise, blank, conversion efficiency, dilution, sample handling and long-run drift. Replicate injection spread addresses only some of these. A laboratory can repeatedly obtain almost the same biased result; accuracy assessment requires independent references, calibration traceability and control checks. The appropriate level of confidence depends on the specific scientific decision, not on how many decimals the software prints.
Reporting limits need separate attention from uncertainty around ordinary measured values. At very low concentrations, background and censoring may dominate. Do not assign a conventional symmetric percentage uncertainty to a below-detection result and then treat it as a precise positive concentration. Use the method’s quality flags and suitable censored-data statistics when summarising nutrient-poor waters. State whether reported concentrations were measured, derived by subtraction, or estimated below quantification.
A fully checked multichannel worked example
Suppose a constructed station table reports nitrate-plus-nitrite 15.20 micromoles/L, nitrite 0.30, reactive phosphate 0.90 and silicate 9.00. Nitrate is therefore 15.20−0.30 = 14.90 micromoles/L, assuming the combined channel has not already been corrected. That arithmetic does not yet tell us what biological process caused the measured composition. It only translates properly defined analytical channels into a nitrate value.
The nitrate-to-phosphate molar concentration ratio is 14.90/0.90 = about 16.6. If a researcher instead includes nitrate plus nitrite as the nitrogen numerator, the corresponding ratio is 15.20/0.90 = about 16.9. Neither ratio becomes biologically meaningful merely because it lies near a familiar number. The reader must identify which measured nitrogen species entered the comparison and why.
Now assume seawater density of 1.025 kg/L for the exercise. Nitrate expressed on a mass basis becomes 14.90/1.025 = about 14.54 micromoles/kg; phosphate becomes about 0.878, and silicate about 8.78. When both numerator and denominator of the nitrate-to-phosphate ratio are converted consistently, the density factor cancels and their molar ratio stays approximately 16.6. Converting only one variable would corrupt the ratio without any change in the water.
How to audit a suspiciously attractive dataset
Begin with the original method and variable dictionary. Identify sample fraction, analytical species, data level, units, reporting limits and applied corrections. Then check a few arbitrary records manually against their calibrations and metadata. Plot nutrient concentration against run order as well as depth. A smooth vertical profile can still conceal a drift pattern if samples were analysed in a consistent sequence, so the two graphs address distinct possible errors.
Next inspect check standards, reference samples, blanks and replicate variability across the run. Flag samples near detection limits, after unusual dilution, after very concentrated samples or around documented maintenance. Compare the resulting pattern with temperature and salinity, but resist altering valid outliers just to make them fit a simple model. An awkward yet reproducible datum may contain the most important natural signal in the entire profile.
Finally write a conclusion whose strength follows the evidence. A defensible sentence identifies quality-controlled bottle nutrient concentrations in stated units and the physical layer where they were collected. It may describe a correlation with oxygen or a water-mass feature, but it should distinguish this observation from a unique causal mechanism. A conclusion about biological limitation or anthropogenic input requires additional tests or context, not a dramatic colour scale.
Common misreadings and the repairs they require
One frequent error is calling a nitrate-plus-nitrite column ‘nitrate’ while retaining nitrite as an additional independent pool, then adding both to obtain ‘total inorganic nitrogen’. That double-counts nitrite. The repair is to write explicit chemical definitions and construct the combined total consistently. If ammonium is available, the analyst can describe a broader dissolved inorganic nitrogen sum, but should not silently include unmeasured components.
Another mistake is comparing phosphate concentrations expressed as micromoles per litre with nitrate concentrations expressed as milligrams of nitrogen per litre and calculating a ratio as if the raw numbers had common units. Any dimensionless molar ratio requires both chemical quantities to be converted into comparable amounts. The apparent ratio might look familiar by coincidence, but dimensional analysis reveals the error. Write the conversion and species name beside each term.
A third mistake is smoothing away below-limit observations and interpolating precise values between sparse depth samples. This manufactures spatial detail that the method never measured. Preserve quality flags and sample levels, disclose the interpolation method where used, and interpret uncertain features conservatively. Visual polish should not replace a data-integrity argument. A trustworthy graph lets readers see where evidence is thin.
Independent practice with checked answers
Exercise one: nitrate-plus-nitrite is 7.40 micromoles/L and separate nitrite is 0.60 micromoles/L. Answer: nitrate is 6.80 micromoles/L when the first channel includes both species. If the delivered column has already made the subtraction, use it as provided. Correct species accounting is as important as correct subtraction.
Exercise two: calibration uses A = 0.025 + 0.100 C and a sample absorbance of 0.225. Answer: C = (0.225−0.025)/0.100 = 2.00 micromoles/L. Dividing by slope alone would include background as nutrient. Check that this sample falls within the validated standard range and matrix before accepting the inversion.
Exercise three: a nutrient-rich sample is diluted to one-fifth its original concentration. The measured diluted value is 3.5 micromoles/L. Answer: the undiluted concentration is 17.5 micromoles/L, provided instrument processing has not already applied the dilution factor. The exact aliquot and final volume should be recorded, not merely described as ‘diluted five times’.
Exercise four: a phosphate signal gives 0.03 micromoles/L but the method’s quantification limit is 0.05. Answer: the displayed number can be retained as an estimate under a documented convention, but it should not be announced as precise quantitative phosphate. ‘Below quantification’ differs from chemical absence and from a missing sample. Preserve the applicable QC code.
Exercise five: density is 1.025 kilograms/L and nitrate is 10.25 micromoles/L. Answer: nitrate is 10.00 micromoles/kg. Dividing by 1,025 rather than 1.025 would confuse kilograms per cubic metre with kilograms per litre and shift the result by three orders of magnitude. The unit next to the density matters.
Exercise six: nitrate is 0.80 micromoles/L and phosphate 0.020. Answer: their concentration quotient is 40 on a molar basis, but near-limit phosphate uncertainty may make the ratio unstable. The quotient alone does not establish phosphorus limitation. Report concentrations, detection limits and biological context before interpretation.
Exercise seven: a low surface sample follows a deep sample with much greater nutrient concentration, and the surface peak is unexpectedly high. Answer: check carryover, wash performance, run sequence and repeat aliquots. An actual oceanographic feature is possible, but so is tubing memory. A nearby station’s shape is not enough to adjudicate the laboratory question.
Exercise eight: nitrate from a float and a ship bottle differs by 2 micromoles/kg. Answer: the discrepancy is not automatically a sensor error. Compare pressure, sampling time, water mass, adjusted versus provisional float fields and shipboard references. Different observation systems and collocation uncertainty must be examined before a performance claim.
Exercise nine: a figure labels 0.000 as ‘no phosphate’. Answer: find the original data value and flag. Zero may be a rounded estimate, a detection-limit code, an invalid sample or an actual numerical substitution. Rewrite the caption to report the method’s detection status, not an unsupported assertion of absolute absence.
Exercise ten: a sample contains 14 micromoles of nitrate ion per litre and is reported in mg N/L. Answer: multiply 14 micromoles by about 14.01 micrograms N per micromole, giving about 0.196 mg N/L. Multiplying by nitrate ion’s molar mass would instead express mg nitrate/L, a different metric.
Exercise eleven: a cruise shows low phosphate and high chlorophyll fluorescence. Answer: this is consistent with several hypotheses, including biological nutrient drawdown, physical supply history or fluorescence-specific behaviour. Compare time, light, pigment assays, other nutrients and mixed-layer conditions. Correlation and a plausible story do not by themselves prove a mechanism.
Exercise twelve: a laboratory changed calibration standards midway through a multiyear archive. Answer: check intercomparison data and the history of detection limits and processing. A discontinuity may be instrumental rather than environmental. A transparent analysis can show the raw series and the effects of any validated harmonisation, without disguising uncertainty as certainty.
Exercise thirteen: a large silicate concentration appears only in one channel after reagent replacement. Answer: examine reagent composition, chemical interference controls, blank, calibration and instrument log. Silicate is not inferred directly from a large optical peak without assay validation. Repeat or qualify suspect measurements and preserve the record of the suspected cause.
Exercise fourteen: two surveys both say ‘nitrate’, but one includes nitrite and the other subtracts it. Answer: their labels refer to different measured or derived quantities. If separate nitrite measurements are available, harmonise to an explicitly chosen definition while retaining provenance. If not, describe the limitation instead of pretending the datasets are chemically identical.
Questions readers most often ask
Can an autoanalyser measure every form of nutrient? No. Its reagent channels detect defined chemical species or operational fractions. Organic forms, particles and other nitrogen or phosphorus compounds may require distinct methods. Read the full analyte definition before expanding a channel result into a claim about a total element pool.
Is the Redfield ratio a pass-or-fail test for healthy seawater? No. It is a useful stoichiometric reference, not an ecological verdict. The concentrations of measured dissolved nutrients, their supply and uptake rates, organism requirements and other resources all affect interpretation. The ratio must remain subordinate to the full evidence.
Do below-detection results mean zero? No. They mean the assay could not reliably establish a quantity under the stated threshold and conditions. Data systems may retain estimated signals or censor them. Avoid replacing that uncertainty with a categorical claim of absolute absence.
Why do density and units matter if seawater is nearly one kilogram per litre? ‘Nearly’ can still matter when comparing precise measurements, and some mistakes involve a factor of one thousand rather than a small density correction. The safe practice is to write the dimensional conversion once and apply it consistently.
What should accompany a nutrient profile? At minimum station, time, depth or pressure, species definition, units, sample handling, quality flags and a stated analytical method. Temperature, salinity, dissolved oxygen, light and other chemical observations help make the environmental interpretation more credible.
Where can I learn the related measurements? The ocean CTD guide explains physical structure, and the ocean oxygen guide addresses another biogeochemical response. Neither should be substituted for a direct nutrient assay.
Original methodological references
NOAA NCEI hydrographic nutrient chemistry and sample analysis description.
NOAA NCEI World Ocean Atlas nutrient climatologies for regional context.
Argo measurement and quality guidelines and biogeochemical float program.