A large research synthesis, published in one of the world’s most influential scientific journals, has detected a decline in the amount of dissolved oxygen in oceans around the world — a long-predicted result of climate change that could have severe consequences for marine organisms if it continues.
The paper, published Wednesday in the journal Nature by oceanographer Sunke Schmidtko and two colleagues from the GEOMAR Helmholtz Centre for Ocean Research in Kiel, Germany, found a decline of more than 2 percent in ocean oxygen content worldwide between 1960 and 2010. The loss, however, showed up in some ocean basins more than others. The largest overall volume of oxygen was lost in the largest ocean — the Pacific — but as a percentage, the decline was sharpest in the Arctic Ocean, a region facing Earth’s most stark climate change.
The loss of ocean oxygen “has been assumed from models, and there have been lots of regional analysis that have shown local decline, but it has never been shown on the global scale, and never for the deep ocean,” said Schmidtko, who conducted the research with Lothar Stramma and Martin Visbeck, also of GEOMAR.
Ocean oxygen is vital to marine organisms, but also very delicate — unlike in the atmosphere, where gases mix together thoroughly, in the ocean that is far harder to accomplish, Schmidtko explained. Moreover, he added, just 1 percent of all the Earth’s available oxygen mixes into the ocean; the vast majority remains in the air.
Climate change models predict the oceans will lose oxygen because of several factors. Most obvious is simply that warmer water holds less dissolved gases, including oxygen. “It’s the same reason we keep our sparkling drinks pretty cold,” Schmidtko said.
But another factor is the growing stratification of ocean waters. Oxygen enters the ocean at its surface, from the atmosphere and from the photosynthetic activity of marine microorganisms. But as that upper layer warms up, the oxygen-rich waters are less likely to mix down into cooler layers of the ocean because the warm waters are less dense and do not sink as readily.
“When the upper ocean warms, less water gets down deep, and so therefore, the oxygen supply to the deep ocean is shut down or significantly reduced,” Schmidtko said.
The new study represents a synthesis of literally “millions” of separate ocean measurements over time, according to GEOMAR. The authors then used interpolation techniques for areas of the ocean where they lacked measurements.
The resulting study attributes less than 15 percent of the total oxygen loss to sheer warmer temperatures, which create less solubility. The rest was attributed to other factors, such as a lack of mixing.
Matthew Long, an oceanographer from the National Center for Atmospheric Research who has published on ocean oxygen loss, said he considers the new results “robust” and a “major advance in synthesizing observations to examine oxygen trends on a global scale.”
Long was not involved in the current work, but his research had previously demonstrated that ocean oxygen loss was expected to occur and that it should soon be possible to demonstrate that in the real world through measurements, despite the complexities involved in studying the global ocean and deducing trends about it.
That’s just what the new study has done.
“Natural variations have obscured our ability to definitively detect this signal in observations,” Long said in an email. “In this study, however, Schmidtko et al. synthesize all available observations to show a global-scale decline in oxygen that conforms to the patterns we expect from human-driven climate warming. They do not make a definitive attribution statement, but the data are consistent with and strongly suggestive of human-driven warming as a root cause of the oxygen decline.
“It is alarming to see this signal begin to emerge clearly in the observational data,” he added.
“Schmidtko and colleagues’ findings should ring yet more alarm bells about the consequences of global warming,” added Denis Gilbert, a researcher with the Maurice Lamontagne Institute at Fisheries and Oceans Canada in Quebec, in an accompanying commentary on the study also published in Nature.
Because oxygen in the global ocean is not evenly distributed, the 2 percent overall decline means there is a much larger decline in some areas of the ocean than others.
Moreover, the ocean already contains so-called oxygen minimum zones, generally found in the middle depths. The great fear is that their expansion upward, into habitats where fish and other organism thrive, will reduce the available habitat for marine organisms.
In shallower waters, meanwhile, the development of ocean “hypoxic” areas, or so-called “dead zones,” may also be influenced in part by declining oxygen content overall.
On top of all of that, declining ocean oxygen can also worsen global warming in a feedback loop. In or near low oxygen areas of the oceans, microorganisms tend to produce nitrous oxide, a greenhouse gas, Gilbert writes. Thus the new study “implies that production rates and efflux to the atmosphere of nitrous oxide … will probably have increased.”
The new study underscores once again that some of the most profound consequences of climate change are occurring in the oceans, rather than on land. In recent years, incursions of warm ocean water have caused large die-offs of coral reefs, and in some cases, kelp forests as well. Meanwhile, warmer oceans have also begun to destabilize glaciers in Greenland and Antarctica, and as they melt, these glaciers freshen the ocean waters and potentially change the nature of their circulation.
When it comes to ocean deoxygenation, as climate change continues, this trend should also increase — studies suggest a loss of up to 7 percent of the ocean’s oxygen by 2100. At the end of the current paper, the researchers are blunt about the consequences of a continuing loss of oceanic oxygen.
“Far-reaching implications for marine ecosystems and fisheries can be expected,” they write.
Extended Data Figures
- Extended Data Figure 1: Dissolved oxygen and apparent oxygen utilization changes per decade since 1960. (446 KB)
- a, Change of dissolved oxygen (DO) per square metre per decade (in units of percentage of local dissolved oxygen); these data are similar to those in Fig. 1. b, Change of apparent oxygen utilization (AOU) in units of mol per square metre per decade.
- Extended Data Figure 2: Oxygen solubility changes. (270 KB)
- a–c, Zonal upper 2,500 m mean oxygen solubility changes in the Atlantic (a), Indian (b) and Pacific (c) oceans. No substantial changes are observed below 1,000 m. Contour lines represent oxygen concentrations at 20 μmol kg−1 and every 30 μmol kg−1.
- Extended Data Figure 3: Temperature and salinity changes. (261 KB)
- a–f, Zonal mean temperature (T; a–c) and salinity (S; d–f) changes in the Atlantic (top row), Indian (middle row) and Pacific (bottom row) oceans per decade. Data locations and handling are identical to those used for the oxygen computation and for the results in Extended Data Fig. 1. This is only a small subset of data available for global temperature and salinity trend computations. Contour lines represent the mean fields. Observed trends are similar to trends described in the literature and thus confirm that no artificial trend is created because of sparse or irregular data locations or through the mapping method. Distortions in the North Atlantic around 40° N are due to the Mediterranean Sea, which creates a discontinuity in the zonal means.
- Extended Data Figure 4: Oxygen concentration and changes. (309 KB)
- a–f, Zonal mean oxygen concentrations in the Atlantic, Indian and Pacific oceans (a–c, respectively), and respective changes in oxygen concentration per decade (d–f). Contour lines are at 20 μmol kg−1 and every 30 μmol kg−1.
- Extended Data Figure 5: Oxygen profile data coverage since 1900. (637 KB)
- Blue indicates locations of oxygen profiles over 5-year data intervals, given at the top of each panel.
- Extended Data Figure 6: Time span of observations and time of last observation in data sets. (465 KB)
- a–j, Time span in years is colour coded (key at right of each panel); each panel shows results for the indicated depth layer. k–t, Year of last observation of data for each grid point mean state and trend computation is colour coded; each panel shows the results for a particular depth layer.
- Extended Data Figure 7: Expected oxygen loss distribution from artificial bias. (321 KB)
- a–c, Zonal mean changes in dissolved oxygen (colour key at right) of an induced systematic bias of 0.5% for historic measurements (Methods) for Atlantic (a), Indian (b) and Pacific (c) oceans. Note the different order of magnitude of colour scales compared to Extended Data Fig. 4. Solid lines represent the mean oxygen field, as in Extended Data Fig. 4.
- Extended Data Figure 8: Trend of zonal oxygen loss using reduced data sets. (259 KB)
- a–f, Change in dissolved oxygen of reduced oxygen data distributions for 20,000 profiles per decade (a–c) and 30,000 profiles per decade (d–f), validating the robustness of the mapping with ‘strongly reduced’ and ‘reduced’ data sets in comparison with the full data set as presented in this Letter (see Methods for details). The global mean trends related to these maps are 946 ± 526 Tmol per decade (a–c) and 988 ± 459 Tmol per decade (d–f). Solid lines represent the mean oxygen field, as in Extended Data Fig. 4.