Response to ‘The importance of accurate CO₂ dosing and measurement in ocean acidification studies’

In his comment on our paper, Moran (Moran, 2014) raises the important issue of ensuring that experiments investigating the responses of marine organisms to future ocean acidification scenarios are carried out with a high degree of certainty about the CO₂ levels being tested. We agree wholeheartedly, which is why we take considerable care in measuring and validating pCO₂ in our field-based and laboratory experiments. Here we explain, more fully than was possible in a Short Communication, the procedures used to measure and cross-validate pCO₂ in Chung et al. (Chung et al., 2014) and other studies we have conducted over recent years (in which the methods have been reported). These techniques are in accordance with those mentioned by Moran and detailed descriptions are already available in the literature (e.g. Hari et al., 2008). We also correct a number of factual errors and incorrect assumptions made by Moran in his comment.

Moran states that because of the ‘dependence of the Chung et al. study on pH to determine test pCO₂ levels, the dose–response effect reported in the paper should be treated with caution’ and that ‘we cannot assert that the target CO₂ levels were correct’. The author proposes that: ‘Ideally, direct pCO₂ measurements should be made via nondispersive infrared measurement (NDIR), where the dissolved gases are equilibrated with a carrier gas which passes through an infrared analyser.’ Indeed, that is exactly what was done in our experiment. While we did not have room in Chung et al. (Chung et al., 2014) to report on the method, we have previously reported that we cross-validate our estimates of pCO₂ from seawater carbonate chemistry using NDIR (e.g. Munday et al., 2010; Munday et al., 2013; Simpson et al., 2011; Watson et al., 2014). As described by Hari et al. (Hari et al., 2008), we used a Vaisala GMP343 infrared CO₂ probe (accuracy ±5 ppm CO₂ + 2% of reading over the range of our experimental manipulations) in a closed loop to measure xCO₂ in the experimental treatment. Temperature-, pressure- and humidity-corrected pCO₂ by NDIR was 946±14 μatm (mean ± s.e.m., N=9), which is within 2 μatm of the value estimated by seawater chemistry (944±19 μatm). This confirms the reliability of our methods and the accuracy of the pCO₂ treatment reported in the paper.

Moran mistakenly refers to our current-day control (466±15 μatm CO₂) as a near-future treatment. It is well known that pCO₂ of seawater in coral reef lagoons is elevated above atmospheric levels in summer because of enhanced calcification by corals and the reduced flushing of lagoons. For example, the daily average pCO₂ on a Hawaiian coral reef was 431–622 μatm (Shamberger et al., 2011). This is the environment inhabited by coral reef fish, and thus the appropriate current-day control. It is not a near-future treatment.

Moran incorrectly states that we kept fish for days in the same body of water, which could lead to significant organic loading that would affect the estimates of total alkalinity (TA) used to calculate pCO₂. As in our previous studies, fish were kept in replicate 30 l aquaria in a flow-through system, and the paper clearly states that each aquarium received a continuous flow of CO₂-equilibrated seawater at 1 l min⁻¹. A turnover time of 30 min is sufficient to flush dissolved metabolic waste from the tanks. Tanks were regularly cleaned of any solid waste. The close match between our estimates of pCO₂ from seawater chemistry and NDIR confirms that loading of non-carbonate buffers was not an issue in our experiment. Nevertheless, we agree with Moran that organic loading is an important issue to consider when designing experiments to study the biological effects of ocean acidification. Finally, as we report in our paper, we validated our measured values of TA against the relevant seawater standards. As we have reported elsewhere, our laboratory measures TA within 1% of the reference value.

The choice of methods to characterize seawater carbonate chemistry depends on the parameters of interest, the precision and accuracy required for studies of biological effects versus detecting ongoing changes in ocean chemistry, and other aspects of experimental design. Like many of our studies, Chung et al. (Chung et al., 2014) was conducted at a remote field station on the Great Barrier Reef, which imposes certain logistical constraints on the methodology. Spectrophotometry is a highly accurate method to measure pH; however, it is not practical to transport and operate a high-precision spectrophotometer at a remote field station. Similarly, it is often not practical to rely on dissolved inorganic carbon and TA estimations for replicated biological experiments at field stations because of the expense involved in transporting and analysing large numbers of seawater samples from numerous replicate tanks, and the impractical lag-time before results are returned to confirm experimental treatments. Provided is it carefully conducted, and preferably cross-validated with other techniques as we have done in our studies, pH can be a useful parameter for controlling experimental treatments and for characterizing seawater chemistry in biological experiments.

Moran correctly points out that glass pH electrodes may drift when continually immersed in seawater. That is one of the reasons we take daily measurements in the experimental tanks, so that we can cross-calibrate the dosing pH probes with the realised pH in our experiments and then adjust the pH set point of the dosing system as required. For seawater chemistry, we take daily, or twice daily, readings of pH in each aquarium using a high-quality pH meter and laboratory-grade electrode (Mettler Toledo). A fault of some other studies is that they rely on the continuously measured pH at the site where pCO₂ is manipulated, not in the tanks where the animals are being treated. We agree that has the potential to introduce errors in the reported seawater parameters, because of both electrode drift and CO₂ flux between the location of CO₂ dosing and the experimental tanks.

One advantage of using glass electrodes and careful calibration with NBS buffers, and cross-validation with NDIR, is that we can have large, appropriately replicated, experiments with multiple levels of CO₂ treatment. While it is important to ensure accurate seawater chemistry in ocean acidification experiments, there are many other issues of experimental design that can cause far greater bias in our understanding of the biological effects of environmental stressors. Inadequate replication, unrealistically high treatment levels, and insufficient duration of experiments are serious problems in many ocean acidification studies. We argue that, more so than chemistry methods that are already well established, this is an area that needs to be improved in ocean acidification research.

Toward the end of his comment, Moran proposes that one method of ensuring accurate test pCO₂ is to equilibrate water with a known gas composition derived from a gas mixing pump, mass flow mixer or pre-purchased gas cylinder. While this method can be successful for small volumes of static water that are continuously aerated, it can be difficult to achieve full equilibration in larger volumes of seawater, especially in a flow-through system. Good equilibration can be achieved with counter-current towers or purpose-designed diffuser membranes, but good equilibration is difficult without specialist equipment that would be difficult to deploy in a field-based situation. We warn the reader that this method is more challenging than it appears at face value, especially for experiments that require large volumes of flowing water, such as those with fish and other metazoans. One approach we have found successful is to first achieve the desired pCO₂ by treating to a set pH in separate header tanks or sumps and then aerating the experimental tanks with the desired gas composition.

Moran makes some useful comments for new researchers entering the field of ocean acidification research who may not be aware of the difficulties and challenges of accurately estimating seawater carbon parameters, including pCO₂. Perhaps the most important point, however, is the final comment about the need to accurately predict the effects of climate change. While achieving well-constrained chemistry in ocean acidification studies is part of that goal, we cannot hope to reliably predict the future impacts of ocean acidification on complex marine ecosystem without addressing the most challenging and important knowledge gaps in the field. Major knowledge gaps include the capacity for long-term acclimation and adaptation to ocean acidification (Sunday et al., 2014), interactive effects with other stressors, and how impacts on individual organisms scale up to affect ecological processes and ecosystem function (Hilmi et al., 2013). Having the vision to tackle these crucial questions, as some laboratories around the world are starting to do, is what will really drive ocean acidification research forward.