Effects of ocean acidification on Antarctic marine organisms: A meta‐analysis

Abstract Southern Ocean waters are among the most vulnerable to ocean acidification. The projected increase in the CO2 level will cause changes in carbonate chemistry that are likely to be damaging to organisms inhabiting these waters. A meta‐analysis was undertaken to examine the vulnerability of Antarctic marine biota occupying waters south of 60°S to ocean acidification. This meta‐analysis showed that ocean acidification negatively affects autotrophic organisms, mainly phytoplankton, at CO2 levels above 1,000 μatm and invertebrates above 1,500 μatm, but positively affects bacterial abundance. The sensitivity of phytoplankton to ocean acidification was influenced by the experimental procedure used. Natural, mixed communities were more sensitive than single species in culture and showed a decline in chlorophyll a concentration, productivity, and photosynthetic health, as well as a shift in community composition at CO2 levels above 1,000 μatm. Invertebrates showed reduced fertilization rates and increased occurrence of larval abnormalities, as well as decreased calcification rates and increased shell dissolution with any increase in CO2 level above 1,500 μatm. Assessment of the vulnerability of fish and macroalgae to ocean acidification was limited by the number of studies available. Overall, this analysis indicates that many marine organisms in the Southern Ocean are likely to be susceptible to ocean acidification and thereby likely to change their contribution to ecosystem services in the future. Further studies are required to address the poor spatial coverage, lack of community or ecosystem‐level studies, and the largely unknown potential for organisms to acclimate and/or adapt to the changing conditions.

: Database search strategy with main themes (top line) and synonyms below. Searches were conducted using 'and' between columns and 'or' between rows for each column e.g. ocean or marine and acid* or pH and "Southern Ocean". For Scopus database " " were replaced with { }. ocean acid* "Southern Ocean" marine pH Antarctic* pelagic ?CO2 coast* DIC nearshore "carbon dioxide" "dissolved inorganic carbonate chemistry" "carbonate chemistry" Strobel, A., Bennecke, S., Leo, E., Mintenbeck, K., Pörtner, H. O., and Mark, F. C. (2012). Metabolic shifts in the Antarctic fish Notothenia rossii in response to rising temperature and pCO 2 . Frontiers in Zoology, 9:28.
-Not included as carbonate chemistry measurements required to convert pH (measured in the paper) to CO 2 (µatm) not provided in the paper and the corresponding author could not be contacted.
-Not included as this is a transcriptomic study and does not have usable measurements (mean and standard error) to input into meta-analytic statistical analyses.
-Not included as this was a sea-ice study with a high ambient CO 2 treatment (1651 µatm).
-Not included as this is a genomic study and does not have usable measurements (mean and standard error) to input into meta-analytic statistical analyses.  , Hellessey, N., Kawaguchi, S., Nicol, S., Hoem, N., and Virtue, P. (2018). Adult Antarctic krill proves resilient in a simulated high CO 2 ocean. Nature Communications, 1(1):190.
-Not included as this is a long-term study (46 weeks) investigating the acclimative capacity of Antarctic krill to ocean acidification.
-Not included as this was a multistressor experiment investigating the effect of ocean acidification and iron availability. There was no ambient iron treatment only iron enriched and iron deplete where hydroxamate siderophore desferrioxamine was added to bind and reduce the bioavailability of iron.
-Not included as this is a transcriptomic study and does not have usable measurements (mean and standard error) to input into meta-analytic statistical analyses.
Huth, T., and Place, S. (2016). Transcriptome wide analysis reveal a sustained cellular stress response in the gill tissue of Trematomus bernacchii after acclimation to multiple stressors. BMC Genomics, 17:127.
-Not included as this is a transcriptomic study and does not have usable measurements (mean and standard error) to input into meta-analytic statistical analyses.  Figure S1: Forest plots of all bacterial response ratios and variance included in the metaanalysis. The data is separated by CO 2 treatment at which the response was measured and information is provided on the study paper, experiment and biological measurement from which the response ratio is calculated. At the end of each CO 2 bracket summary statistics from weighted, random effects are provided including the Q statistic, degrees of freedom, p-value, I 2 , mean response ratio and 95% confidence interval.   Figure S2: Forest plot of all phytoplankton response ratios, 95% confidence intervals and variance (v) included in the meta-analysis.   Figure S2: Forest plot of all phytoplankton response ratios, 95% confidence intervals and variance (v) included in the meta-analysis.    Figure S2: Forest plot of all phytoplankton response ratios, 95% confidence intervals and variance (v) included in the meta-analysis (A: pre-industrial and 500-800 µatm, B: 801-1000 µatm, C: 1001-1500, 1501-2000 and >2000 µatm). The data is separated by CO 2 level at which the response was measured and information is provided on the study paper and biological response from which the response ratio is calculated. At the end of each CO 2 level summary statistics from weighted random effects models are provided including the Q statistic, degrees of freedom, p-value, I 2 , mean response ratio and 95% confidence interval. Summary statistics are also provided at the bottom of the figure for all phytoplankton response ratios with all CO 2 levels together.  Figure S3: Forest plots of all macroalgal response ratios and variance included in the metaanalysis. The data is separated by CO 2 treatment at which the response was measured and information is provided on the study paper, species and biological measurement from which the response ratio is calculated. At the end of each CO 2 bracket summary statistics from weighted, random effects are provided including the Q statistic, degrees of freedom, p-value, I 2 , mean response ratio and 95% confidence interval.  Figure S4: Forest plots of all invertebrate response ratios and variance included in the metaanalysis. The data is separated by CO 2 treatment at which the response was measured and information is provided on the study paper, organism and biological measurement from which the response ratio is calculated. At the end of each CO 2 bracket summary statistics from weighted, random effects are provided including the Q statistic, degrees of freedom, p-value, I 2 , mean response ratio and 95% confidence interval.    Figure S5: Forest plots of all fish response ratios and variance included in the meta-analysis. The data is separated by CO 2 treatment at which the response was measured and information is provided on the study paper, species and biological measurement from which the response ratio is calculated. At the end of each CO 2 bracket summary statistics from weighted, random effects are provided including the Q statistic, degrees of freedom, p-value, I 2 , mean response ratio and 95% confidence interval.