The global oxygen budget and its future projection.

Atmospheric oxygen (O2) is the most crucial element on earth for the aerobic organisms that depend on it to release energy from carbon-based macromolecules. This is the first study to systematically analyze the global O2 budget and its changes over the past 100 years. It is found that anthropogenic fossil fuel combustion is the largest contributor to the current O2 deficit, which consumed 2.0 Gt/a in 1900 and has increased to 38.2 Gt/a by 2015. Under the Representative Concentration Pathways (RCPs) RCP8.5 scenario, approximately 100Gt (gigatonnes) of O2 would be removed from the atmosphere per year until 2100, and the O2 concentration will decrease from its current level of 20.946% to 20.825%. Human activities have caused irreversible decline of atmospheric O2. It is time to take actions to promote O2 production and reduce O2 consumption.


Introduction
O 2 is the most crucial atmospheric component for lives on earth, which is maintained not only by the process of photosynthesis by green plants and algae but also the processes that consume O 2 , such as respiration, combustion and decomposition [1]. Observations [2] have revealed that with the rapid development of industrialization and modern civilization, the concentration of atmospheric O 2 has been declining over the past 30 years. Simultaneously, the O 2 levels in oceans have also been decreasing due to the change of solubility under the back ground of global warming [3], and more dead zones have appeared [4].
Comparing to the rapid increase of CO 2 concentration and its climate impacts, the decline of atmospheric O 2 is far beyond the focus of research community and policy makers due to its negligible changes compared to its massive inventory in the Earth's atmosphere. In fact, the decline in atmospheric O 2 should be much more addressed [5] since it could affect the survival of humans and most of the species directly. Here, based on observations [6] and Fifth Coupled Model Intercomparison Project (CMIP5) simulations [7], this study diagnoses the global O 2 budget systematically to provide a clear understanding of O 2 decline.

Data and methods
In this section, some important issue involved in our research is discussed, including definitions of several terms commonly used in atmospheric O 2 work, the method of estimating the consumption and production of O 2 and the construction of global O 2 cycle.

The observational oxygen concentration data
Typically, the concentrations of gas are reported in the unit of volume fraction (e.g., ppm, ppb, etc.). However, the concentration of atmospheric O 2 are reported as changes in the O 2 /N 2 ratio of air relative to a reference (air collected in the mid-1980s) to avoid the non-negligible interference caused by dilution effects. The observed changes are very small and are reported in per meg units [8]: where the subscripts ''sample" and ''reference" indicate the sample air and the reference air, respectively. The changes observed in the O 2 /N 2 ratio are very small. One per meg equals 10 À4 percent, or 10 À6 . Observational O 2 concentration data of nine stations around the world from the Scripps O 2 Program (http://scrippso2.ucsd.edu/) is used in this study. These data are from remote locations or other locations situated so that they represent averages over large portions of the globe rather than local background sources [6].

The estimation of oxygen consumption
O 2 is consumed by a wide range of processes, including (1) autotrophic respiration, (2) heterotrophic and soil respiration, (3) fires, (4) fossil fuel combustion and industry, (5) the weathering of organic matter and sulfide minerals, and (6) volcanic gas oxidation [9]. The main cause of the O 2 decrease in the atmosphere is fossil fuel combustion. Population growth and the growing number of livestock, which directly impacts human livelihoods, also contribute to the depletion of atmospheric O 2 by heterotrophic respiration. In addition, deforestation, tropical peatland fires, and the burning of agricultural waste not only contribute to the increase in atmospheric CO 2 but also remove a significant amount of O 2 from the atmosphere. Here we mainly discuss the following four processes, since the other processes are either hard to quantify or tiny enough to be neglected. All the data are gridded to a 1.0°Â 1.0°resolution for analysis.

Oxygen consumption by fossil fuel combustion
The estimation of O 2 consumption by fossil fuel combustion is based on CO 2 emissions data from the Carbon Dioxide Information Analysis Center (CDIAC, http://cdiac.ess-dive.lbl.gov/). According to Keeling [9], about 1.4 mol of O 2 is consumed when 1 mol of CO 2 are emitted. For future projections of O 2 combustion, the global total carbon emission data under RCP4.5 and RCP8.5 from 2005 to 2100 is obtained from RCP scenario data group (http://www.pikpotsdam.de/~mmalte/rcps/).

Human respiration
O 2 consumption by human respiration is based on the population density datasets from the Gridded Population of the World, Version 4 (GPWv4, http://sedac.ciesin.columbia.edu/). The population counts for the future scenario (SSP1 and SSP3) are provided by Murakami et al. [10]. We assume that an adult at rest consumes approximately 21 L of O 2 per hour and in a day, a man works 8 h with a labor intensity between light and medium (1.0 L O 2 /min) and rests (21 L O 2 /h) for the remaining 16 h. According to the standard above, an adult consumes approximately 1.17 kg (816 L) of O 2 per day.

Livestock consumption
O 2 consumption by livestock respiration is based on the spatial distributions of main livestock from Gridded Livestock of the World v2.09 [11]. The basal metabolism rate (BMR) is the rate of energy expenditure per unit time by endothermic animals at rest and can be reported in mL O 2 /min. The BMR (mL O 2 /h) of a mammal can be predicted with the formula given by Kleiber [12], BMR = 3.43 M 0.75 , where M is the animal's mass (g). Following this formula, the annual O 2 consumption of the livestock can be estimated (Table 1). In the future projections and historical simulations, we assume that the total number of all livestock is proportional to the total human population.

Fire
O 2 consumption by fire is based on the data on carbon emissions from fire activities derived from the Global Fire Emissions Database (GFED, http://www.globalfiredata.org) [13]. The GFED combines satellite information on fire activity and vegetation productivity to estimate gridded monthly burned area and fire emissions as well as scalars that can be used to calculate higher-temporal resolution emissions. The current version of this dataset is version 4, which has a spatial resolution of 0.25°and ranges from 1997 to 2016. O 2 consumption by fire is estimated assuming that the O 2 :CO 2 molar ratio is 1.1. The consumption of O 2 by fire changes little annually, and we regard this value as constant (5.87 Gt/a) in the future scenarios and historical simulations.

Oxygen production by land
O 2 is produced during the processes of photosynthesis, in which the plants and other organisms absorb carbon dioxide (CO 2 ) from the atmosphere and release oxygen (O 2 ). The photosynthesis can be expressed by the following chemical equation: Gross primary production (GPP) is the total amount of CO 2 fixed by a plant in photosynthesis. Net Primary Productivity (NPP) is the net amount of gross primary productivity remaining after including the costs of plant respiration [14][15][16]. The remaining fixed energy is referred to as net primary productivity (NPP). Net Ecosystem Productivity (NEP) refers to the net amount of primary productivity remaining after including the costs of respiration by plants, heterotrophs, and decomposers. Therefore, NEP = GPP -(R a + R h + R d ), where R a is the autotrophic respiration, R h is the respiration by heterotrophs and R d is the respiration by decomposers (microbes). A measure of NEP is of great interest when determining the CO 2 balance between various ecosystems, even the entire Earth, and the atmosphere. The O 2 balance is closely linked to the CO 2 balance.
According to Eq. (2), we can use the following equation to calculate the net amount of O 2 produced during the processes of photosynthesis with the known net carbon fixed (NEP). The unit of O 2 production is g m À2 a À1 . In this paper, the simulated NEP dataset from 1900 to 2100 is obtained from the simulation by CMIP5 models (Table 2) and are gridded to 1.0°Â 1.0°r esolution for analysis. Some models directly provide NEP while others provide NPP and R h .

Air-sea oxygen flux
Ocean is another important source of atmospheric O 2 . The CMIP5 models (Table 3) provide monthly mean air-sea O 2 flux in the unit of mol m À2 s À1 . In this study, we convert the unit to g m À2 a À1 and grid the data to 1.0°Â 1.0°resolution for analysis.

The oxygen budgets
The processes that release O 2 to the atmosphere (e.g., photosynthesis) and the processes that consume O 2 (e.g., respiration, fires, fossil fuel combustion, the weathering of organic matter, and volcanic oxidation) result in large fluxes of O 2 to and from the atmosphere and constitute the global O 2 cycle [1]. A slight disturbance in production or consumption can generate large shifts in atmo-spheric O 2 concentrations. Based on the discussion of production and human-related O 2 consumption in the previous sections, the global O 2 cycle is constructed.
where D ATM is the rate of decline in global atmospheric O 2 concentrations; C FF , C RES , C FIRE is the consumption of fossil fuel, humans and livestock and fire respectively. P LAND and O OCEAN represent the production from land and outgassing from the ocean. The equation above omits the respiration of wild animals, weathering of organic matter and volcanic oxidation, which are insignificant compared to the processes above and are hard to quantify. Thus, the residual term is introduced to correct this bias and is calculated based on the difference between the observational D ATM and the simulated D ATM from 1991 to 2005. All terms above are reported in Gt/a.

Results analysis
The four main processes including fossil fuel combustion, human and land livestock respiration, and fires, are presented in   [13], but also permanently reduces the global production of O 2 by photosynthesis, thus causing accelerating O 2 depletion. The O 2 production over land could be quantified by the net ecosystem production (NEP), and the climatological distribution of NEP from CMIP5 simulation is presented in Fig. 2a. It shows that total amount of NEP is 5.28 Gt/a (equivalent to 14.08 Gt/a of O 2 ) and 72.2% is provided by the tropics. Under the RCP4.5 and RCP8.5 scenarios, the O 2 production from land rises to 16.75 Gt/a and 19.44 Gt/a, respectively, by the end of the 21st century, and the most rapid increase occurs in the tropics (Fig. 2b and c), especially in Central Africa and Southeastern Asia. The changes of NEP are mainly determined by the NPP (net primary production) variability, which is easier to be measured. Under climate change, the global NPP presents an increasing trend and the reason could be attributed to the following three aspects. Firstly, the increase of atmospheric CO 2 has a positive effect on NPP because atmospheric carbon is a driving factor for the photosynthesis of C 3 plants [17]. Secondly, nitrogen deposition can increase the biomass in nitrogen-limited northern temperate forests and result in an increase of the NPP [18,19]. Thirdly, global warming leads to the lengthening of the plant growing season [20] and the increasing of precipitation [21], which also exert positive effects on the increasing NPP. However, the O 2 increase caused by the above processes cannot compensate for the O 2 consumption by humans' activities on land. If fossil fuel combustion is not limited, relying only on the self-adjustment of terrestrial ecosystems will not make much difference in maintaining the atmospheric O 2 concentration.
The ocean is the second O 2 library except for the continent. Fig. 3a presents the CMIP5 simulated climatological distribution of the oceanic O 2 flux, which shows a net influx from oceans to the atmosphere at low latitudes and the opposite occurring at high latitudes, with a global total outgassing of 1.6 Gt per year. The projections of O 2 flux under these two scenarios differ in magnitude but follow remarkably similar trends overall (Fig. 3b and c). The global O 2 flux will experience increases of 1.2-2.7 Gt/a during the 21st century under RCP4.5 and RCP8.5, respectively, based on CMIP5 models. Although the flux increases under both scenarios, this does not mean that more O 2 is produced by marine plants.
In fact, the significant decrease in NPP indicates that the ocean O 2 production is reduced and the marine environment is experiencing deterioration [22]. Models show that most of the world's oceans are suffering from NPP reduction, including areas where oceanic O 2 outgassing has increased. The increasing O 2 flux may be attributed to the changes of solubility, ocean circulation and convection. An increase in ocean temperature leads to a decrease in solubility and stratifies the ocean, thus limiting ventilation and the supply of O 2 to the interior [23][24][25], causing more O 2 to be outgassed from oceans to the atmosphere.   respectively. The residual term, which includes the systematic bias, is about 2.69 Gt. In total, the O 2 depletion in the atmosphere is 21.23 Gt/a, which is mainly associated with the growth rate of atmospheric CO 2 concentration. Fig. 5a shows the temporal variations of each term of the O 2 budget from 1900 to 2100 (with the period of 1990-2005 by historical simulations and 2006-2100 by RCP8.5 projections). The O 2 production over land has increased from 5.97 to 17.43 Gt/a, and the fossil fuel combustion has increased from 1.99 to 29.76 Gt/a during 1900-2005. This indicates that the enhancement of photosynthesis rate is not significant compared with the rapidly rising anthropogenic O 2 consumption under the background of global warming. The accelerated increasing fossil fuel combustion is the dominant factor which leads to the widening of the gap between O 2 consumption and production, and then results in the accelerated depletion of atmospheric O 2 . By the projections under RCP8.5, this difference between consumption and production would be extended. A significant decrease of O 2 appears throughout the whole century, and approximately 100 Gt of O 2 would be removed from the atmosphere each year by the end of the 21st century (Fig. 5b). The O 2 concentration would decrease from its current level of 20.946% to 20.825% (RCP8.5) and 20.89% (RCP4.5) by the end of the 21st century.

Conclusion and discussion
The above results indicate that the decreasing trend of atmosphere O 2 is significant, which has been much neglected by the public. Here we emphasize that the current O 2 that has accumulated in the atmosphere and dissolved in the oceans throughout a billionyear Earth history is not limitless. This O 2 inventory is strongly threatened by humans' aggressive activities. Increasing amounts of O 2 are being consumed by increasing fossil fuel combustion along with population growth, and accelerated deforestation [26];  moreover, the expansion of drylands [27] will also reduce the O 2 production of terrestrial ecosystems. The O 2 in the ocean also faces severe threaten. Marine garbage has emerged as a serious problem [28] and the number of dead zones on Earth has doubled every decade since the 1960s [4]; these factors have limited the O 2 production in oceans and caused waters to lose O 2 . The ''deoxygenation" and expansion of O 2 -minimum zones (OMZs) in oceans indicate the arrival of hypoxia in marine ecosystems. These hidden risks associated with the ocean O 2 crisis are directly related to the O 2 inventory on Earth. All of the cumulative effects described above that limit the output of O 2 are putting humanity's future at risk. It is foreseeable that life on Earth will inevitably suffer from hypoxia in the future if we continue these extravagant activities.
Thus, to save our earth, we must take more immediate actions to promote the output of O 2 and reduce its consumption, such as by using more green energy instead of combusting more fossil fuels, recycling more municipal and industrial trash on land [29], and using more anaerobic microorganisms to decompose organic matter [30], such that the rate of O 2 decline can be decelerated. It is also pivotal to reverse this trend through the combined efforts and cooperation of all countries; otherwise, the human race, as well as other aerobes, will be left behind forever, and our dominance of this planet will become just a brief footnote in its long history [5]. We are entering a new era in Earth's history in which humans, rather than natural forces, are the primary drivers of planetary change. Instead of further degradation, we can redefine our relationship with Earth from a wasteful, unsustainable and predatory one to one where people and nature can coexist in harmony.