Chung et al. (Chung et al., 2014) recently reported the results of a study investigating the effect of dissolved CO2 level on the retinal response of a marine fish. The two CO2 concentrations tested represented different ocean acidification scenarios. The lower concentration represented a near-future CO2 concentration (466 μatm), and the other a 2100 scenario with elevated CO2 due to anthropogenic inputs (944 μatm). Such studies are timely given the high probability that atmospheric CO2 concentration will continue to increase. The results will likely be incorporated into future Intergovernmental Panel on Climate Change reports, and may form part of future action plans designed to preserve species and ecosystem viability. For this reason, it is crucial to ensure that any testing of future ocean acidification scenarios is carried out with a high degree of certainty about the CO2 levels tested.
The methods Chung et al. used to both quantify and dose CO2 in their study lack the necessary accuracy and precision to definitively assert that the target CO2 test levels were correct. The key problem is their dependence on pH measurements made with a glass potentiometric electrode and low ionic concentration calibration buffers (a method commonly employed by a number of the authors on this paper), which is not recommended best practice for the accurate determination of pH in waters of high ionic concentration (Dickson, 2010), such as seawater. A cross-laboratory comparison of the accuracy of glass pH electrodes by Illingworth (Illingworth, 1981) found a mean measurement error of 0.2 pH units for high ionic concentration solutions, with many probes performing worse. Recalibration does not help, as many electrodes will apparently calibrate perfectly well in the low ionic concentration NBS (National Bureau of Standards) buffers (probably the most commonly used pH calibration buffers), but when used in saltwater will give erroneous measurements. Sources of error for seawater pH measurements based on glass electrodes and NBS buffer calibration include sodium ion error (e.g. Na+ is detected as H+), differences in ionic diffusivity between the reference electrode solution and the sample solution, and the fact that NBS reference buffers are not directly related to hydrogen ion concentration (which is necessary to compute dissolved CO2) (Dickson, 1993; Grasshoff, 1983; Millero et al., 1993). Small errors in pH measurement will have significant effects on the computed dissolved CO2 concentration. For example, for seawater (34.5 ppK salinity, 30°C, total alkalinity 2269 μmol kg−1 SW) with pH 7.9, the computed pCO2 is 604 μatm, and at pH 7.8 pCO2 is 788 μatm. For reference, the range of CO2 concentrations typically used in ocean acidification scenarios (400 versus 1000 μatm) are equivalent to a pH difference of between 0.3 and 0.4 pH units. In addition, when glass potentiometric electrodes are continually submersed in saltwater, they quickly drift and lose measurement span, limiting the value of pH meters as a method to control the dosing of pure CO2 into experimental tanks. In light of the known limitations glass electrodes present for measuring seawater pH, and the total dependence of the Chung et al. study on pH to determine test pCO2 levels, the dose–response effect reported in their paper should be treated with caution.
Researchers and manuscript reviewers need to be aware of the difficulty of accurately measuring and dosing CO2 when approaching studies such as that of Chung et al. (Chung et al., 2014). Best practice guidelines for ocean acidification research have been developed (Riebesell et al., 2010), and there is plenty of information available on methods for accurate dissolved CO2 measurement (Dickson et al., 2007). In order to ensure ocean acidification studies of aquatic biota are valid and repeatable, we must ensure that the test CO2 concentrations are both accurate and precise. The solution is to adopt the measurement methods used by ocean chemists, which not only rely on techniques with a high degree of measurement certainty, but are also subject to quality control and assessment (i.e. the use of reference material to validate measurements) (Dickson et al., 2007). Ideally, direct pCO2 measurements should be made via nondispersive infrared measurement (NDIR), where dissolved gases are equilibrated with a carrier gas which then passes through an infrared analyser (the Alliance for Coastal Technologies has evaluated some of the main manufacturers of this technology, www.act-us.info). If pH is to be used for pCO2 calculation (together with either total inorganic carbon or alkalinity), then it is necessary to have an absolute measurement of H+ concentration that is free of the errors introduced by potentiometric glass electrodes and non-specific calibration buffers. The most accurate and precise pH measurement method is spectrophotometry plus an indicator dye (Millero et al., 1993), which can be automated to monitor seawater pH continuously (McGraw et al., 2010). Total alkalinity (TA) can be measured with a high degree of precision using potentiometric electrodes and commercially available acid standard solutions, and together with pH, TA can be used to calculate pCO2 [as was done by Chung et al. (Chung et al., 2014)]. However, there is an assumption that the components contributing to the alkalinity are known and accounted for, which allows for the calculation of HCO3− and CO32− concentration ([HCO3−]+[CO32−]=TA–[non-carbonate buffers]). In oceanic seawater with a low organic loading, the carbonate alkalinity represents almost all of the TA, but in waters with a significant organic loading, the contribution of non-carbonate buffers to total alkalinity may be substantial. In experiments where fish are kept for days in the same body of water [as is common in experiments such as that of Chung et al. (Chung et al., 2014)], the contribution of metabolic waste products (ammonia, nitrates, phosphates, organic acids, fatty acids and proteins) to alkalinity and pH balance increases significantly. As CO2 is a small percentage of total inorganic carbon (at atmospheric CO2 concentrations), small errors in HCO3− + CO32− measurement will result in substantial pCO2 calculation error. For waters with complex alkalinity compositions, total inorganic carbon is a better carbonate parameter to measure (via water sample acidification plus NDIR analysis of carrier gas CO2 concentration). One method of ensuring accurate test pCO2 levels is to equilibrate water with a known gas composition, which can be derived from a gas-mixing pump, mass flow mixer or pre-purchased from a gas cylinder supplier.
For good reasons, ocean acidification research and its possible biological impacts are targeted areas for funding and are of considerable public interest. The failure to carry out accurate CO2 dose–response experiments will limit our ability to find general patterns between ocean acidification studies. It can potentially waste research resources as the results may be uninterpretable, and there is the real possibility that the media, public and stakeholders will lose confidence in our societal value if we fail to predict the effects of climate change accurately.