Recommendations with respect to the impact of identified problems with the
measurement of carbonate system parameters on the UK-OA (Ocean Acidification)
research programme
Cause for concern

Increased inputs of CO2 to the atmosphere result in increased in the amounts of
CO2 dissolved in seawater. CO2 reacts with seawater releasing hydrogen ions. The
water is more acidic and the concentration of carbonate ions is decreased [1].
Consequently solid calcium carbonate (shells) become more soluble [2] and the
balances of other biogeochemical processes are shifted.

Much research into ocean acidification aims to predict shifts in biogeochemical
processes and organism function. A particular concern is damage to organisms
with shells formed of calcium carbonate. The concentration of carbonate in
solution needs to be accurately determined. Concerns have been raised that some
methods used to do this give disparate results [3].
Desk study evidence

Carbonate ion concentration cannot be measured directly. Several carbonate
system components (including carbonate ion concentration) have to be calculated
from measurements of any two of the four directly measurable carbonate system
quantities: total alkalinity (TA), total dissolved inorganic carbon (DIC), partial
pressure of carbon dioxide in solution (pCO2) and hydrogen ion (as pH). The
calculations use well established chemical equilibrium data.

TA and DIC can potentially be measured accurately and precisely in a range of
different water sample types and sample sizes. pCO2 can be measured accurately
and precisely at sea using underway monitoring systems but an accepted system
for smaller scale laboratory measurements is lacking. New direct measurements of
pH are potentially more reliable than older data because suitable certified
reference solutions are now available to control the accuracy of measurements. [4]

Calculations of the carbonate system assume that carbonate alkalinity is a known
fraction of the measured TA [3], this may not be true in some waters relevant to the
study of ocean acidification. Organic alkalinity – organic matter that reacts with
the acid used in the measurement of alkalinity by titration - is not accounted for in
the calculations.

Otherwise no problems exist with the choice of software package for the
calculation of carbonate system variables. They produce near-identical results
when given the same input data to calculate the rest of the carbonate system.

Numerically different sets of ‘constants’ for use in these calculations have been
developed. Discrepancies in the output produced when different sets of constants
are used increases as the pCO2 of the system increases. At a future level of 1000
µatm [5], the maximum calculated discrepancy is 5 %.

Deviations of up to 25 % have been reported between direct measurements of
pCO2 and calculations of pCO2 based on measured values of TA and DIC. The
quality of the measurements on which this claim was based was questioned in
peer reviews of the work.

Evidence is emerging that in experimental systems and coastal waters, inclusion
of organic alkalinity in the measured total alkalinity can produce large (10s of %)
discrepancies in calculated values of pH, carbonate ion concentration and pCO2
when calculated from measurements of TA and DIC.

As far as is currently known, errors are only significant in artificially manipulated
waters (e.g. culture experiments, tank experiments and mesocosms) and inshore
waters [6].

The choice of which 2 parameters to measure critically determines the accuracy of
the calculated result. The error in estimation of other carbonate system
components resulting from the presence of organic alkalinity in the measured TA
is most significant when it is paired with DIC. This is shown graphically in the
full report.

When one of the two measured parameters is pH, it is primarily the accuracy of
the pH measurement that determines the accuracy of calculations of other
parameters.
Impact on UKOARP and related studies

Monitoring to be carried out in Area A of the UKOARP is concerned with open
sea and ocean waters, where these concerns have potentially little impact.

Monitoring carried out in the DEFRA-pH project does include waters in which the
potential problem has been identified.

UKOARP cruises (e.g. Area B) should look to over-determine the carbonate
chemistry (at least 3 parameters measured) when not in open ocean waters.

Many of the studies in UKOARP are based on culture experiments. Measurements
of TA in these systems are likely to be effected by the presence of dissolved
organic matter.

The plan was for the UKOARP Analysis Service to determine the carbonate
system by measurements of TA and DIC. Where organic alkalinity is significant
this approach will lead to inaccuracies.
Recommendations UKOARP

The community working in OA research has to be made aware of the potential
errors in the measurement of TA identified in recent research.

All work supported by UKOARP must provide an assessment of error in the data
they report for the determination of pCO2 and pH in their systems.

Established best practice for measurements should be followed [7,8].

Meta-data describing the details of the experiments, analytical methods and their
quality should accompany all data transferred into databases or published reports
and papers. A template for doing this is provided in the main report.

The measurements made in DEFRApH of TA, DIC and pCO2 need to be
rigorously compared to determine if any deviations are consistent with an error
induced by the presence of organic alkalinity in the samples.

The UKOARP Analytical Service needs to measure a third parameter (probably
pH), if the accuracy of carbonate system determinations is to be ensured.
Wider Recommendations
A cause for concern has been raised over uncertainty in the calculation of
carbonate system variables particularly under future conditions. A possible cause
for the problem has been identified, however the evidence base is small therefore
supporting studies are required.

The experiments carried out by Hoppe et al
[9]
laboratories using agreed “best practice” methods
should be repeated in other
[7]
. This would determine if
indeed the currently used equations for calculations of carbonate system variables
are not appropriate for use at artificially high concentrations of CO2.

The limited studies that have identified the presence of titrate-able organic matter
[10]
should be repeated in a wider range of waters to determine in exactly which
waters the presence of organic matter contributes to the measured value of total
alkalinity.

Measurements of alkalinity have been widely used because certified reference
solutions were available to assess the accuracy of the measurements and high
precision measurements can be made relatively easily. Reference solutions have
recently become available for use in the direct measurement of pH. Operating
procedures should be developed to assist the wider community to make high
quality measurements of pH
[4]
. This will make it easier to cross check analyses
and calculations of concentrations of carbonate system components.

Development of techniques to measure pH accurately, reliably and cheaply would
greatly facilitate OA research.
Footnotes and supporting information
[1] Increased inputs of CO2 to the atmosphere result in an increase in the amount of
CO2 dissolving in seawater. What happens next is governed by series of chemical
equilibrium reacts that readjust the composition of seawater at the ionic level to take
account of the newly introduced CO2. The concentration of CO2 - gas dissolved is the
water increased and the CO2 reacts with the water to form carbonic acid. This
carbonic acid in turn rapidly dissociates into its component ions H+ (hydrogen ions),
HCO3- (bicarbonate) and CO32- (carbonate) and then a slower adjustment takes with
respect to solid carbonate compounds in contact with the water - in particular this
will be with the shells of organisms which are formed of calcium carbonate (CaCO3).
Increasing the amount of CO2 in seawater actually decreases the amount of
carbonate (CO3-) in solution thereby increases the tendency for the shells of
organisms to dissolve, as the system adjusts to a new equilibrium state. This is first
potential problem associated with ocean acidification. The second is that water
actually becomes more acidic due to the increase in hydrogen ions in the waters
(acidity is measure of the quantity hydrogen ions in solution - is reported on the pH
scale - the negative logarithm to base 10 of the hydrogen ion concentration so that a
smaller value is equivalent to higher acidity). This change in acidity can cause shifts
other chemical equilibria such as in the concentration of ammonium changing the
fertility of the water.
[2] To assess how increased levels of CO2 in the atmosphere will change the viability
carbonate shelled organisms requires that the “saturation state” of water is known (a
measure of the tendency of solid calcium carbonate (shells) to dissolve). This requires
that the carbonate concentration in waters of interest is known. These measurements
are usually made indirectly – by calculation following a measurement of the total
alkalinity in the sample. This requires that the equations used for this calculation are
accurate and applicable to all the waters of interest in studies of ocean acidification.
There is considerable evidence that the equations are good in ocean waters but there
is now a small amount of evidence that this might not be true in coastal waters and
experimental systems containing high concentrations of organic matter and at
concentrations of CO2 beyond currently normal ranges.
[3] Measurement and calculation: Concentrations of components carbonate system
are either measured directly or calculated from measurements of two of the four most
easily determined measures of the system. The choice of measurement pair critically
determines the accuracy of the calculated result. See report sections 4 and 5 and
Zeebe, R. E. and Wolf-Gladrow, D. A.:2001 CO2 in Seawater: Equilibrium, Kinetics,
Isotopes, Elsevier Science, Amsterdam, Netherlands.
[4] To assist with the development of high quality pH measurements outside the core
community of carbonate chemists a more closely specified standard operating
procedure (SOP) than the current CO2-SOP-6a (Dickson et al., 2007) for the
electrode based measurements of pH is recommended – this would be based on
experience already available in the community on the most appropriate pH electrodes
and temperature sensors to use and the appropriate design of a measurement cell.
Development of instruments for the automated colorimetric determination of pH
should be encouraged as these offer the possibility of making high precision
measurements (down to < 0.001 pH unit) with small (< 5 ml) volumes of sample).
[5] A worst case scenario for increased concentrations of CO2 will occur by 2010
(http://www.ipcc-data.org/ddc_co2.html)
[6] The table below illustrates the propagation of the error to estimations pCO2 and
pH from the potential errors in measured TA that have been identified. In these
calculations the other variable in the calculation DIC, pCO2 and pH was assumed to
be without error. The results here should be compared to the error maps in section 4
of the main report.
Reported errors in measurements of TA and the propagated errors in calculated values
of other carbonate system variables - pCO2 and pH around pCO2 levels of 380 and
1000 μatm.
Source
Report
of error
Total
TA
C µM/l error
µM/l
pCO2 (DIC)
pCO2
pH
380/1000
(pH)
(pCO2)
decrease
increase
increase
μatm
μatm
PIC
Shelf sea [1]
8
<16
26/101
3/7
0.003/0.003
Plankton
Cultures [2]
~ 200
<5
9/33
1/2
0.001/0.001
Bacteria
Coastal [2]
<30
<6
10/40
1/3
0.001/0.001
DOC
Cultures [3]
80
80
107/386
14/36
0.013/0.014
DOC
Cultures [4]
<800
338/927
136/351
0.112/0.120
DOC
Coastal [4]
<200
199/646
34/88
0.032/0.034
DOC
Mesocosm [5]
20
32/124
3/9
0.003/0.003
TA - total alkalinity; DIC - total dissolved inorganic carbon, PIC - particulate inorganic carbon; DOC dissolved organic carbon
380/1000 calculations performed to give final pCO2 concentrations of 380 amd 1000 ppm
Errors on direct measurements TA ± 2, DIC ± 2, pCO2 ± 2, pH ± 0.005 and annual rate of increase in
pCO2 (+2 pH -0.002)
[1] Harlay, J., A.V. Borges, C. Van Der Zee, B. Delille, R.H.M. Godoi, L.-S. Schiettecatte, N. Roevros,
K. Aerts, P.-E. Lapernat, L. Rebreanu, S. Groom, M.-H. Daro, R. Van Grieken, L. Chou
(2010) Biogeochemical study of a coccolithophore bloom in the northern Bay of Biscay (NE
Atlantic Ocean) in June 2004. Progress in Oceanography 86, 317-336.
[2] Kim, H. C., K. Lee, and W. Y. Choi (2006), Contribution of phytoplankton and bacterial cells to the
measured alkalinity of seawater, Limnol. Oceanogr., 51, 331–338.
[3]Kim, H.-C., and K. Lee , 2009. Significant contribution of dissolved organic matter to seawater
alkalinity, Geophys. Res. Lett., 36, L20603, doi:10.1029/2009GL040271.
[4] Hernandez-Ayon, J. M., Zirino, A., Dickson, A. G., Camiro-Vargas, T., and Valenzuela-Espinoza,
E., 2007. Estimating the contribution of organic bases from microalgae to the titration
alkalinity in coastal seawaters. Limnology and Oceanography: Methods 5, 225-232.
Hernández-Ayón, J. M., Belli, S. L., and Zirino, A., 1999. pH, alkalinity and total CO2 in
coastal seawater by potentiometric titration with a difference derivative readout. Analytica
Chimica Acta 394, 101-108.
[5] Muller, F. L. L. and Bleie, B., 2008. Estimating the organic acid contribution to coastal seawater
alkalinity by potentiometric titrations in a closed cell. Analytica Chimica Acta 619, 183-191.
[7] Best practice is well documented by the EPOCA and IOCCP projects - Gattuso, J.P., Lee, K., Rost, B., and Schulz, K.: 2010. Approaches and tools to manipulate the carbonate
chemistry, Guide for Best Practices in Ocean Acidification Research and Data 10 Reporting,
Office for Official Publications of the European Union, Luxembourg.
Dickson, A. G., Sabine, C. L., and Christian, J. R.: 2007. Guide to best practices for ocean
CO2 measurements, PICES Special Publication, 3, Sidney, Canada, 2007.
In addition to existing best practice – developments are needed appropriate to working in
experimental systems:- (1) A SOP appropriate to sampling experimental systems is
needed; this would include recommendations on the collection of small volume
samples (< 100 ml) and filtering of samples. Research is needed into reliable and
easy to use (by non expert) storage containers particularly for the storage of small
samples (10 to 100 ml). (2) Research is needed to find a “safe” efficient biocide to
replace the use of mercuric chloride. Note: The testing of biocides and containers
must be done in ways that the results are statistically valid.
[8] Specific recommendations for ensuring the quality of measurements of total alkalinity are
given in section 3.5 of the main report.
[9] Hoppe, C. J. M., G. Langer, S. D. Rokitta, D. A. Wolf-Gladrow, and B. Rost, 2010. On
CO2 perturbation experiments: over-determination of carbonate chemistry reveals
inconsistencies. Biogeosciences Discuss., 7, 1707–1726.
[10] Kim, H.-C., and K. Lee , 2009. Significant contribution of dissolved organic matter to
seawater alkalinity, Geophys. Res. Lett., 36, L20603, doi:10.1029/2009GL040271.
Hernandez-Ayon, J. M., Zirino, A., Dickson, A. G., Camiro-Vargas, T., and ValenzuelaEspinoza, E., 2007. Estimating the contribution of organic bases from microalgae to the
titration alkalinity in coastal seawaters. Limnology and Oceanography: Methods 5, 225-232
Muller, F. L. L. and Bleie, B., 2008. Estimating the organic acid contribution to coastal
seawater alkalinity by potentiometric titrations in a closed cell. Analytica Chimica Acta 619,
183-191.