Clumped isotopes in near surface atmospheric CO 2 over land , coast and ocean in 1 Taiwan and its vicinity 2

Abstract. Molecules containing two rare isotopes (e.g., 13C18O16O in CO2), called clumped isotopes, in atmospheric CO2 are powerful tools to provide an alternative way to independently constrain the sources of CO2 in the atmosphere because of their unique physical and chemical properties. We presented clumped isotope data (Δ47) in near-surface atmospheric CO2 from urban, suburban, ocean, coast, high mountain ( ∼  3.2 km a.s.l.) and forest in Taiwan and its vicinity. The primary goal of the study was to use the unique Δ47 signature in atmospheric CO2 to show the extents of its deviations from thermodynamic equilibrium due to different processes such as photosynthesis, respiration and local anthropogenic emissions, which the commonly used tracers such as δ13C and δ18O cannot provide. We also explored the potential of Δ47 to identify/quantify the contribution of CO2 from various sources. Atmospheric CO2 over ocean was found to be in thermodynamic equilibrium with the surrounding surface sea water. Respired CO2 was also in close thermodynamic equilibrium at ambient air temperature. In contrast, photosynthetic activity result in significant deviation in Δ47 values from that expected thermodynamically. The disequilibrium could be a consequence of kinetic effects associated with the diffusion of CO2 in and out of the leaf stomata. We observed that δ18O and Δ47 do not vary similarly when photosynthesis was involved unlike simple water–CO2 exchange. Additionally we obtained Δ47 values of car exhaust CO2 that were significantly lower than the atmospheric CO2 but higher than that expected at the combustion temperature. In urban and suburban regions, the Δ47 values were found to be lower than the thermodynamic equilibrium values at the ambient temperature, suggesting contributions from local combustion emission.


Introduction
The budget of atmospheric CO 2 is widely studied using the temporal and spatial variations in concentration and conventional isotopic compositions (δ 13 C and δ 18 O) of CO 2 (Francey and Tans, 1987;Francey et al., 1995;Yakir and Wang, 1996;Ciais et al., 1995aCiais et al., , b, 1997;;Peylin et al., 1999;Cuntz et al., 2003;Drake et al., 2011;Welp et al., 2011;Affek and Yakir, 2014).δ 13 C is useful to differentiate the exchange of CO 2 with the ocean and land biospheres.This is due to the fact that the photosynthetic discrimination against 13 C during exchange with land plants is higher than that associated with the chemical dissolution of CO 2 in the ocean (e.g., Tans et al., 1993;Ciais et al., 1995a;Francey et al., 1995;Ito, 2003;Bowling et al., 2014).The major limitation of δ 13 C is that it cannot distinguish CO 2 produced by high-temperature combustion or low-temperature respiration (Affek and Eiler, 2006;Laskar et al., 2016a).δ 18 O in atmospheric CO 2 is mainly controlled by various water reservoirs (ocean, leaf and soil).In urban locations, a significant fraction of CO 2 may have combustion origin possessing δ 18 O signature of atmospheric O 2 (Kroopnick and Craig, 1972;Ciais et al., 1997;Yakir and Wang, 1996).δ 18 O is used for partitioning net CO 2 terrestrial fluxes between soil respiration and that exchange with plant leaves, the exchange rate is enhanced by the presence of carbonic anhydrase in plants and soils (Francey and Tans, 1987;Farquhar and Lioyd, 1993;Yakir and Wang, 1996;Ciais et al., 1997;Peylin et al., 1999;Murayama et al., 2010;Welp et al., 2011).This is because δ 18 O of CO 2 fluxes originated from soil respiration is different from that exchanged with leaf water.δ 18 O in soil water reflects the δ 18 O value of local meteoric water while leaf water is relatively enriched due to transpiration.The δ 18 O values from these processes and interactions are different and hence the tracer is widely used for constraining the gross production of CO 2 (Francey and Tans, 1987;Ciais et al., 1997;Gillon and Yakir, 2001;Cuntz et al., 2003;Welp et al., 2011).However, due to rapid exchange of oxygen isotopes between CO 2 and different water reservoirs with diverse δ 18 O and processes such as evapotranspiration complicate its interpretation (Riley et al., 2003).
The doubly substituted isotopologues or clumped isotopes such as 13 C 18 O 16 O in CO 2 , denoted by 47 , provides an additional and independent constraint to study the atmospheric CO 2 budget.47 in air CO 2 can help to identify the mechanisms for CO 2 production and consumption.Unlike conventional isotopes, clumped isotope studies for the atmospheric CO 2 are very limited mainly because of challenges to apply it to the atmospheric study (Eiler and Schauble, 2004;Affek et al., 2007;Yeung et al., 2009).The available data are not sufficient to address some key issues such as quantification of CO 2 from different sources and to what extent the air CO 2 is in thermodynamic equilibrium with leaf and surface waters, especially in regions with strong anthropogenic activities such as urban areas.Also the effect of photosynthesis on the 47 of air CO 2 has not been studied rigorously.δ 18 O and 47 were reported to have similar timescales for the isotope exchange between CO 2 and water (Affek, 2013;Clog et al., 2015), but no comparative study on their behavior in presence of other processes such as photosynthesis and respiration was done.A combined assessment from all the three aforementioned isotopic tracers can better constrain the budget of CO 2 and associated processes than δ 13 C or δ 18 O alone.
Theoretically it is shown that in thermodynamic equilibrium, 47 values of CO 2 are temperature dependent (Eiler and Schauble, 2004;Wang et al., 2004), verified over a wide range from 10 to 1000 • C (Dennis et al., 2011).Processes that involve CO 2 and liquid water as medium, such as isotopic exchange with ocean water, are expected to have 47 values close to the thermodynamic equilibrium.47 values in ambient air CO 2 should reflect a balance of CO 2 fluxes between biosphere-atmosphere exchange, ocean-atmosphere exchange and emissions from combustion sources.Photosynthesis involves gas-phase diffusion of CO 2 into leaves, fixes about one-third of the CO 2 and returns the rest back to the atmosphere.CO 2 molecules inside a leaf are generally expected to be in thermodynamic equilibrium with leaf water because of presence of enzymatic carbonic anhydrase that greatly enhances the isotopic exchange (Cernusak et al., 2004).47 values of soil respired CO 2 is also not well constrained, though it is believed to be in thermodynamic equilibrium with the soil water (Eiler and Schauble, 2004).
Here, we present clumped and conventional isotope data in near-surface air CO 2 , covering a wide variety of processes and interactions.Air samplings were made in South China Sea, two coastal stations in northern Taiwan, an urban traffic street, a suburban location, a forest site, a greenhouse, the top of a high mountain and car exhaust.The study was designed and aimed to show the extents of the deviations of near-surface atmospheric CO 2 from thermodynamic equilibrium with local surface water.Possible influences from other processes such as anthropogenic emission, respiration and photosynthesis on clumped isotopes were explored.

Materials and methods
Stable isotopic compositions of CO 2 including mass 47 were measured using a Finnigan MAT 253 stable gas source isotope ratio mass spectrometer configured to measure ion beams corresponding to m/z 44 through 49.The instrument registers the major ion beams (44, 45 and 46) through resistors 10 8 , 3 × 10 10 and 10 11 Ohm, respectively, and minor ion beams (47, 48 and 49) through 10 12 Ohm.All the measurements were carried out at Research Center for Environmental Changes, Academia Sinica, Taiwan.
Air samples were collected in 2 L flasks and compressed to 2 bar atmospheric pressure using a membrane pump.The flasks, equipped with two stopcocks, were first flushed with the ambient air for ∼ 10 min before starting the sample collection.We then closed the downstream end stopcock, allowed the pressure to build to 2 bar and then isolated by closing the other stopcock.The air pumping for flushing and sampling was carried out through a column packed with magnesium perchlorate to remove moisture.The moisture content was reduced from the ambient value of 70-90 % to less than 1 % relative humidity, checked using a LI-COR infrared gas analyzer (model 840A, LI-COR, USA).See, for example, Liang and Mahata (2015) for more details of air sampling.
To show how photosynthesis and respiration affect the abundances of CO 2 isotopologues and to demonstrate what different information the 47 can give from the other isotopologues, we performed analyses for CO 2 collected in a controlled greenhouse with cemented floor located in the top (third) floor of the Greenhouse Building, Academia Sinica.The size of the greenhouse was about 8 m long, 5 m wide and 5 m high.It was closed at least 1 day before each experiment and the ventilation was kept off.More than 70 % of the ground area inside the greenhouse was occupied with Cinnamomum cassia plants, each of ∼ 2 m height kept in pots.Samples were collected at intervals of less than half an hour to a few hours on 3 sunny days and 1 cloudy day to investigate the influence of photosynthesis and respiration on the isotopologues of CO 2 .Inside the room relative humidity was ∼ 50-70 % for the 3 sunny days and was above 90 % for the cloudy day.Forest air CO 2 was collected from a dense natural forest at the west end of the Academia Sinica Campus.The vegetation mainly consists of medium to big size trees with canopy heights varying between 10 and 20 m.The samples were collected ∼ 100 m inside the forest on a small plateau at a height of ∼ 30 m from the ground in the slope of a hill; the dense vegetation allowed little sunlight penetrating to the surface.The relative humidity at the site was 80-90 % during the sampling days and wind speed was nearly zero due to presence of hills on three sides of the sampling spot.Marine air was collected during a cruise in the South China Sea (for the cruise track see Fig. 1) at a height of ∼ 10 m a.s.l. and from two coastal stations: Keelung (25 • 09 6 N, 121 • 46 22 E) and Fugui Cape (25 • 18 N, 121 • 32 E) (Fig. 1) at a height of ∼ 5 and ∼ 20 m a.s.l., respectively.Sea surface temperatures were measured at the time of sampling.Urban air was collected at a bus stop on Roosevelt Road, a busy street in Taipei.Suburban air was collected from an open roof (∼ 30 m above ground) of Institute of Earth Science Building, Academia Sinica (AS; 25 • 2 41 N, 121 • 36 52 E); grassland air was collected from a grass field in front of the Department of Atmospheric Science, National Taiwan University Campus (NTU; 25 • 1 N, 121 • 30 E), Taipei.In addition, we collected air from the summit of the Hehuan mountain (24 • 8 15 N, 121 • 16 32 E; 3.2 km a.s.l.) (Fig. 1) on 9 October 2013.All air samplings were made when there was no rain to avoid direct interaction with the rainwater.Car exhaust was collected from a Mazda 3000 cc TRIBUTE and a Mitsubishi 2400 cc New Outlander, using evacuated 2 L glass flasks from ∼ 20 cm inside the exhaust pipes through a column of magnesium perchlorate.
CO 2 was extracted from air using a glass vacuum line connected to a turbo molecular pump by cryogenic technique.The vacuum line as well as the sample flask connection assembly including its head space was pumped to high vacuum before starting the CO 2 extraction.Air in the flask was pumped through a series of five coiled traps, with the first two immersed in dry ice-acetone slush (−77 • C) for trace moisture removal followed by three in liquid nitrogen (−196 • C).CO 2 was collected from the traps immersed in liquid nitrogen by repeated freeze-thaw technique at liquid nitrogen and dry ice temperatures for further removal of traces of water (see Mahata et al., 2012;Liang and Mahata, 2015 for details).The air was pumped for 40-45 min at a controlled rate of ∼ 90 mL min −1 using a mass flow controller; the pressure on the post-mass-flow controller side of the vacuum line was ∼ 10 mm of Hg.No measurable isotopic fractionation caused by mass flow controller at this flow rate was observed when checked using several aliquots of air from a high-volume compressed air cylinder (∼ 40 L at 2000 psi).For car exhaust, an aliquot of exhaust air was transferred to a 60 mL bottle and CO 2 was fully extracted cryogenically following the same protocol as discussed above (but with mass flow controller step skipped).
CO 2 was further purified from other condensable species like N 2 O, CH 4 and hydrocarbons by means of gas chromatography (Agilent 6890N, with a 3.0 m × 0.3 cm stainless steel column packed with PorapakQ 80/100 mesh, supplied by Supelco Analytical, Bellefonte, PA, USA) with the column kept at −10 • C. High-purity helium (> 99.9999 % supplied by Air Products and Chemicals, Inc.) at 20 mL min −1 was used as carrier gas.CO 2 was eluted first, followed forthwith by N 2 O, and CH 4 , hydrocarbons and traces of water came out much later.To get an optimized condition for CO 2 , we checked the separation of CO 2 from N 2 O with varying proportions and at various temperatures (25 to −20 • C) and found a temperature of −10 • C at which column separated CO 2 from N 2 O perfectly (see Laskar et al., 2016b, for details).The column was baked at 200 • C for more than 2 h prior to use.The conditioned column is good for purifying three samples.At the end of the day, long baking (8-10 h) was performed.At the initial phase the working gas was taken from a high-purity commercial CO 2 called AS-2 (δ 13 C = −32.54‰ with respect to VPDB and δ 18 O = www.biogeosciences.net/13/5297/2016/Biogeosciences, 13, 5297-5314, 2016  (2006): where R 13 and R 18 (ratios 13 C / 12 C and 18 O / 16 O) are obtained by measuring the conventional masses 44, 45 and 46 in the same CO 2 sample and R 17 is calculated assuming a mass-dependent relation with R 18 given by R 17 = R 17 VSMOW R 18 /R 18 VSMOW λ , where exponent λ = 0.5164 is used for all 47 calculations (Affek and Eiler, 2006).The value of λ varies between 0.516 and 0.523 (Hoag et al., 2005;Barkan and Luz, 2012;Hofmann et al., 2012;Thiemens et al., 2014).The variation in 47 was less than 0.01 ‰ at 25 • C when the exponent was varied over the aforementioned range.This variation was comparable to the measurement uncertainty and hence not considered here; all the calcula-tions were based on λ = 0.5164.47 is obtained by measuring CO 2 with respect to which the isotopes among various CO 2 isotopologues are distributed randomly ( 47 ∼ 0 ‰).Practically, this random distribution is approached by heating CO 2 at 1000 • C for more than 2 h (Eiler and Schauble, 2004;Affek and Eiler, 2006).Measurements were made with a stable ∼ 12 V signal at mass 44, with peak centering, background scanning and pressure balancing before each acquisition started.Each sample was analyzed for 10 acquisitions, 10 cycles each, at an integration time of 8 s; the total analysis time was approximately 2.5 h.Masses 48 and 49 were monitored to check isobaric interferences due to contamination of hydrocarbons (Ghosh et al., 2006).Details about the corrections due to nonlinearity related to 47 measurements in the mass spectrometer and reference frame equation for expressing the measured 47 values in absolute reference frame (ARF) were discussed in Laskar et al. (2016b).
To obtain the temperature from the 47 values, we used the following relation (Dennis et al., 2011):  (Laskar et al., 2016b) and the accuracy in 47 values in terms of temperature, based on CO 2 equilibrated with water at known temperatures were better than 3 • C (see Table S2 in Supplement).
For [CO 2 ] measurements, flasks of volume 350 cc were used.These small flasks were connected in series with the larger flasks used for isotopic measurements.[CO 2 ] was measured using a LI-COR infrared gas analyzer (model 840A, LI-COR, USA) at 4 Hz, smoothed with 20 s moving average.The analyzer was calibrated against a working standard (air compressed in a cylinder) with a nominal [CO 2 ] of 387.7 ppmv and a CO 2 -free N 2 cylinder.The reproducibility of LI-COR was better than 1 ppmv.The working stan-dard was calibrated using a commercial Picarro analyzer (model G1301, Picarro, USA) by a series of NOAA/GMD certified tertiary standards with [CO 2 ] of 369.9, 392.0, 409.2 and 516.3 ppmv, with a precision (1 σ standard deviation) of 0.2 ppmv.The [CO 2 ] in car exhaust was estimated by gravimetric technique using an MKS Baratron gauge.
Ambient temperatures were taken from the nearest governmental weather stations (operated by Central Weather Bureau, Taiwan): Nankang (for AS; station code: C0A9G0; 25

Greenhouse CO 2
Diurnal variation in the concentration and isotopic compositions of CO 2 inside the controlled greenhouse is shown in Fig. 2. The lowest CO 2 concentration [CO 2 ] and highest δ 13 C and δ 18 O values were observed during late morning hours, while highest [CO 2 ] and lowest δ 13 C and δ 18 O values were observed during nighttime and early morning before sunrise (Table 1 and Fig. 2a-c), indicating that respiration and photosynthesis played the major role in controlling the variations of the [CO 2 ] and isotopic compositions.A Keeling plot, a graphical approach plotted between isotopic composition and the inverse of the concentration, is used to determine the isotopic composition of the source (Pataki et al., 2003).It is valid for a mixing of two components; the intercept of the plot gives the source isotopic composition.Respiration was the main source of CO 2 here added to the background CO 2 .Keeling analysis for δ 13 C had an intercept of −26.32 ± 0.40 ‰ (Fig. 2d), a value expected for C 3 plant respired CO 2 .The Keeling plot for δ 18 O had an intercept of 30.68 ± 0.73 ‰ (Fig. 2e), which could be explained by a combined effect of respired CO 2 equilibrated with soil wa-ter and kinetic fractionation associated with the diffusion of CO 2 from soil to the air.A Keeling plot for δ 13 C with the early morning and nighttime greenhouse data, when photosynthesis was absent, was found to have same intercept as observed with all the data, only the correlation was better for the latter (R 2 = 0.999, not shown).The tight correlations among [CO 2 ], δ 13 C and δ 18 O (Fig. 2d-f) suggest that photosynthesis and respiration were the dominant processes controlling their variations, while mixing with ambient air and anthropogenic contribution of CO 2 were insignificant.
In contrast, 47 shows different patterns of diurnal variability due to the effect of photosynthesis and respiration.Figure 3a-d detail diurnal variations in 47 in the greenhouse CO 2 on 4 different days.The first three were bright sunny days with photosynthesis as the dominant process while the last one was a dark cloudy day affected more by respiration.
To further reduce photosynthetic activity on the last day, two layers of black cloth that cut down the incident sunlight by ∼ 50 % were deployed.The measured 47 values were also compared with the thermodynamic equilibrium values.The maximum value of 47 was observed in the morning before ∼ 08:00 and at night and the values were similar to the thermodynamic equilibrium values at the ambient temperatures.This indicates that the respired CO 2 was in close thermodynamic equilibrium with the leaf and soil water.The daytime (from 09:00 to 17:00) 47 values for the 3 sunny days were higher than the thermodynamic equilibrium values.The  morning before ∼ 09:00 and increase afterwards (Fig. 3).By comparing the 47 values acquired on the sunny days with that on the cloudy day, we noticed that when photosynthesis was weak, the 47 value was close to the thermodynamic equilibrium with soil and leaf water (Fig. 4).The correlation between 47 and [CO 2 ], δ 13 C or δ 18 O (Fig. 3d) was observed only when the photosynthesis was weak.This suggests that 47 carries information different from concentration and conventional isotopic composition when photosynthesis occurs.See Sect.4.1 for detailed discussion.

Car exhaust
The [CO 2 ], δ 13 C and δ 18 O values of car exhaust CO 2 were 39 350 ± 50 ppmv, −27.70 ± 0.03 and 25.35 ± 0.07 ‰, respectively (Table 2).δ 13 C value was similar to that reported elsewhere (Newman et al., 2008;Popa et al., 2014), the δ 18 O was slightly higher than the atmospheric O 2 (∼ 23.5 ‰), and the source of O 2 for combustion.The average value of 47 for the exhaust from the two cars was 0.273 ± 0.021 ‰, which gave a temperature of 282 ± 17 • C (Table 2).This temperature is much lower than the fuel combustion temperatures (> 800 • C).The possible reason for higher values of δ 18 O and 47 in the exhaust CO 2 than expected was postcombustion partial exchange with water and other gaseous species, released during combustion, inside the catalytic converter and the exhaust pipe (see discussion in Sect.4.2).

Atmospheric CO 2 over ocean and coasts
Isotopic compositions including 47 values obtained for CO 2 over ocean and coasts are presented in Table 3.The averaged [CO 2 ] over ocean between latitudes 18 • 03 and 21 • 17 N was 395 ± 7 ppmv, and the values of δ 13 C and δ 18 O were −8.43 ± 0.19 and 40.72 ± 0.20 ‰, respectively (Table 3).In the coastal stations, the averaged values of [CO 2 ], δ 13 C and δ 18 O were 397 ± 10 ppmv, −8.48 ± 0.11 and 40.70 ± 0.29 ‰, respectively.Both the [CO 2 ] and δ 13 C values over the ocean and coasts were similar to those observed at Mauna Loa during the sampling period, suggesting little contribution from local/regional anthropogenic sources.The Keeling analysis for δ 13 C gave an intercept of −13.61 ‰ (Fig. 5a) for the air CO 2 collected over the ocean and coasts.δ 18 O of air CO 2 over the ocean was close to the isotopic equilibrium values with the surface sea water at the sea surface temperatures (see Sect. 4.3).The 47 values varied between 0.880 and 0.946 ‰ for the marine and coastal CO 2 (Table 3, Fig. 5b), similar to that predicted at thermodynamic equilibrium at sea surface temperatures (obtained using Eq. 2).Therefore, both δ 18 O and 47 values suggest that the air CO 2 over the ocean was in close thermodynamic equilibrium with the underlying sea water.

Atmospheric CO 2 over land
To show how anthropogenic emission affects the isotopic composition especially the 47 values, we analyzed atmospheric CO 2 samples collected near Roosevelt Road, a busy street in downtown Taipei.The averaged values of [CO 2 ], δ 13 C and δ 18 O obtained were 500 ± 50 ppmv, −11.05 ± 0.90 and 39.32 ± 0.94 ‰, respectively (Table 4).A significantly higher [CO 2 ] and lower δ 13 C and δ 18 O values compared to the marine CO 2 showed signatures of a significant contribution from vehicular emissions.47 values near Roosevelt Road were found to be in the range of 0.754 to 0.833 ‰, with an average of 0.807 ± 0.028 ‰ (Table 4).The values were lower by ∼ 0.15 ‰ compared to the thermodynamic equi-   librium value at 20 • C, the ambient temperature around the sampling time, indicating a significant fraction of CO 2 produced at higher temperatures, i.e., of combustion origin.
In the suburban location (Academia Sinica Campus), [CO 2 ] averaged over 4 months was 410 ± 10 ppmv (Table 4), which was ∼ 15 ppmv higher than that observed over the South China Sea and that at Mauna Loa Observatory during the time of sampling.The higher [CO 2 ] suggests contribution from local anthropogenic emissions.δ 13 C values varied between −7.83 and −10.30‰, with an average of −8.78 ± 0.50 ‰.Keeling analysis for δ 13 C gave an intercept of −26.16 ± 1.58 ‰ (Fig. 6), indicating source of CO 2 from C 3 plant respiration and/or combustion.47 values here varied between 0.853 and 0.972 ‰ (Table 4) with an average of 0.897 ± 0.027 ‰, which were significantly less than the thermodynamic equilibrium values (assuming water bodies had the same temperature as the ambient) (Fig. 7).
The averaged [CO 2 ], δ 13 C and δ 18 O over the grassland (inside National Taiwan University Campus) were 410 ± 33 ppmv, −8.95 ± 0.70 and 39.74 ± 1.00 ‰, respectively.The Keeling plot for δ 13 C gave an intercept of −16.98 ± 1.02 ‰ (Fig. 6), indicating that a significant frac-tion of CO 2 originated from C 4 vegetation.This is not surprising as the CO 2 was sampled over a C 4 -dominated grassland (area: ∼ 50 m × 50 m).Unlike greenhouse CO 2 , no statistically significant correlation between δ 18 O and 1/ [CO 2 ] in air CO 2 in these sites was observed (not shown), probably due to the influence of multiple sources and processes on oxygen isotopes of atmospheric CO 2 .Figure 7c shows the 47 values in air CO 2 over the grassland at National Taiwan University Campus.A large variation in 47 was observed (0.885-0.989 ‰) with an average of 0.937 ± 0.030 ‰.Some of the values were close to the thermodynamic equilibrium while the others deviated significantly.
In a small and dense forest near Academia Sinica Campus (Fig. 1), average values of [CO 2 ], δ 13 C and δ 18 O in air CO 2 were 438 ± 16 ppmv, −9.99 ± 0.50 and 40.39 ± 0.63 ‰, respectively (Table 4) during summer (July-August) of 2015.A significantly higher [CO 2 ] and lower δ 13 C values than the background indicate strong contribution of CO 2 from local respiration.47 values fall in the range of 0.887 to 0.920 ‰, with an average of 0.895 ± 0.012 ‰ (Table 4).The values were similar to that expected at thermodynamic equilibrium (Fig. 7d) except on 11 August, when a significant increase in  4).The lower [CO 2 ] and higher δ 13 C than Mauna Loa suggest photosynthetic uptake, which was also seen at grassland site and inside greenhouse on a few occasions.Here the averaged value of 47 was 0.904 ± 0.009 ‰, slightly less than that expected at the ambient temperature (Table 4).

Discussion
A detailed discussion of the results obtained from different locations is presented below.

Greenhouse air CO 2
To minimize anthropogenic alteration and air mixing/transport and to maximize the variations of CO 2 isotopologues by biological processes, a controlled greenhouse provides an ideal environment.Diurnal variation was observed in [CO 2 ], δ 13 C, δ 18 O (Fig. 2) and 47 (Fig. 3) in the greenhouse.Good correlations between [CO 2 ], δ 13 C and δ 18 O suggest common processes affecting all of them, and they were photosynthesis and respiration.Giving 31 July as an example, we estimated the rates of nighttime respiration and daytime photosynthetic uptake using the conventional isotopic compositions (analysis of 47 is discussed separately below).The dimension of the greenhouse room was 8 × 5 × 5 m (length, width and height).The nighttime respiration rate was then estimated to be about ∼ 10 ppmv h −1 (considering change of [CO 2 ] from 17:30 to 21:30; Fig. 2a), or ∼ 4 × 10 13 molecules cm −2 s −1 .Using simple isotopic mass balance, this increase of [CO 2 ] could be satisfactorily explained assuming C 3 respiration as the main source of CO 2 (δ 13 C ≈ −26 ‰; intercept in Fig. 2d) added to the background (−8.5 ‰).Similarly, the same conclusion could be derived by analyzing δ 18 O considering δ 18 O of respired and background CO 2 of 30.68 ‰ (intercept in Fig. 2e) and ∼ 40 ‰, respectively.Thus, we conclude that the main factor that affected the changes in concentration as well as the isotopic compositions in nighttime was respiration.
The daytime net uptake rate can be estimated by taking the changes from early morning to noontime; the [CO 2 ] reduced by 110 ppmv, δ 13 C increased by 3.46 ‰ and δ 18 O by 2.23 ‰ in about 6 h.We calculated the number of molecules and their changes inside the greenhouse assuming simple gas laws.The estimated net photosynthetic uptake was ∼ 1 × 10 14 molecules cm −2 s −1 assuming constant respiration rate that was observed in the night.The photosynthetic discrimination can be calculated using the Rayleigh distillation model where R o and R are the initial and modified 13 C / 12 C or 18 O / 16 O ratios (due to photosynthetic activity), respectively, f is the fraction of the material left, and α is the fractionation factor.The estimated discrimination in 13 C defined by (α − 1), following Eq.( 3), was −16.5 ‰, which was slightly higher than that expected for C 3 type vegetation (∼ −20 ‰) (Farquhar et al., 1989).For 18 O, in addition to photosynthetic uptake, one has to consider an additional effect due to temperature-dependent water-CO 2 equilibrium fractionation.That is, the process decreases δ 18 O by ∼ 0.2 ‰ for an increase of 1 • C in temperature (Brenninkmeijer et al., 1983); from morning to noontime, the temperature effect reduced δ 18 O by 4.4 ‰.Adding this factor to the observed change in δ 18 O yielded a discrimination factor of −12.0 ‰; the value becomes −7.0 ‰, if this additional temperature-dependence is ignored.The value (−12.0 ‰) observed considering the additional exchange with the soil water was slightly higher than that observed previously (−14.4‰) (Flanagan et al., 1997).Here the δ 13 C and δ 18 O values of the respired components were assumed to be −26 and 30 ‰, respectively (see Sect. 3.1).We assume that ca.one-third of the CO 2 molecules in stomata are fixed photosynthetically and the remaining retrodiffuse back to the atmosphere (Farquhar and Lloid, 1993) implying that the CO 2 -water isotopic exchange rate was ∼ 2 × 10 14 molecules cm −2 s −1 .Also we assume that the CO 2 molecules that enter into the leaf stomata get isotopically equilibrated with the leaf water before diffusing back to the atmosphere.This implies an approximately 8 h of oxygen isotope exchange time for CO 2 in the greenhouse room.As a result, we do not expect that CO 2 reached to complete isotopic equilibrium with the substrate water in a few hours inside the room.47 values in the leftover CO 2 could be used to check the disequilibrium.The respired CO 2 was found to be in thermodynamic equilibrium at the ambient temperature, shown by the 47 values of CO 2 in the early morning and nighttime (Fig. 3a-c) and that collected on a cloudy day with suppressed photosynthetic activity (Fig. 3d).The close thermodynamic equilibrium at reduced photosynthetic condition is also shown in Fig. 4a that deviation in 47 from that expected at ambient temperature is small.On sunny days, the [CO 2 ], δ 13 C and δ 18 O values change by 50-115 ppm, 2-4 and 1.1-2.2‰, respectively, in a time period of ∼ 5 h in the morning (Fig. 2). Figure 3 shows that the 47 values retained the thermodynamic equilibrium values in the morning hours (until 09:00) and then deviate from the thermodynamic equilibrium later of the day.The maximal reduction in the 47 values during these morning hours was ∼ 0.05 ‰ (Fig. 3a-c) which is significant, as this value is much higher than the uncertainty of the measurements.An increase in 47 values after ∼ 09:00 was observed.We attribute these changes in the 47 values of the residual CO 2 to photosynthesis as it is seen when photosynthesis is strong.Also we note that there was no significant correlation/anti-correlation between δ 18 O and 47 when photosynthesis was strong (Fig. 3a-c), but it became significant when the photosynthesis was weak (Fig. 3d).Therefore, the plant photosynthesis decouples 47 and δ 18 O, in contrast to pure water-CO 2 isotopic exchange where the two behave similarly as far as isotopic equilibration is concerned (Affek, 2013;Clog et al., 2015).
Strong influence of photosynthesis on 47 was also reported by Eiler and Schauble (2004).They observed a decrease in the 47 values of the residual CO 2 due to photosynthetic assimilation though the effect observed was different for different species.Here we observed a decrease in 47 value of the residual CO 2 initially (first 2 h) due to photosynthesis similar to that observed by Eiler and Schauble (2004), although it later started increasing in response to the photosynthesis.Photosynthesis as a source of disequilibrium was also shown recently by analyzing the clumped isotopes of O 2 (Yeung et al., 2015).Though enzymatic carbonic anhydrase catalyzes the water-CO 2 isotopic exchange toward equilibrium (Peltier et al., 1995;Cernusak et al., 2004) its activity varies.A large variation in the activity of carbonic anhydrase in different vegetation types (C 3 , C 4 ) or within the same type was noted previously (see Gillon and Yakir, 2001, and references therein).Therefore, the reaction may be incomplete, which is limited by the enzymatic activity inside leaves.Furthermore, a box modeling by Eiler and Schauble (2004) demonstrated that gas diffusion through leaf stomata during photosynthesis fractionates the remaining air CO 2 47 value, deviating it from the thermodynamic equilibrium set by leaf water.Mixing of more than one component can also cause change in 47 when δ 13 C and δ 18 O of the components are different (Affek and Eiler, 2006;Laskar et al., 2016a), but this can easily be ruled out as it was not observed when photosynthesis was not very strong (Fig. 3d).More rigorous investigations with controlled experiments using different plants with diverse carbonic anhydrase activities are needed to resolve the issue.
Considering the discrimination for δ 13 C, δ 18 O and variation in the concentration it is possible to model the observed isotopic profile.The Rayleigh model (Eq.3) in terms of δ notation can approximately be written as δ = δ o + ε × ln(f ), where δ o is the initial δ value, f is the fraction of material left and ε is the enrichment factor.Figure 8a shows the concentration profiles for 31 July 2015 inside the greenhouse.With the calculated discrimination factors (ε) of −16.5 and −12.0 ‰ for δ 13 C and δ 18 O, the modeled isotopic profiles along with actual data are shown in Fig. 8b and c.The model data are generated using Rayleigh fractionation relation.Assuming this relation valid for 47 , a discrimination factor of 0.065 ‰ due to photosynthesis was observed in the morning hours of 31 July 2015.Figure 8d shows the 47 profile for the same day along with the actual observed values.The observed data match well with the model plots.Unlike δs, 47 is not a linear quantity as discussed later, the discrimination factor calculated may slightly change when nonlinearity is taken into account.More data, probably at leaf level, will allow us to estimate the photosynthetic discrimination for 47 .

Car exhaust CO 2
Ideally, the 47 value of car exhaust CO 2 should reflect the temperature of fuel combustion inside the combustion chamber, which is > 800 • C.However, the temperature estimated from 47 was found to be 283 ± 18 • C. It is likely that interaction of the sample CO 2 with the exhaust gases and water inside the catalytic converter and exhaust pipe modified the 47 values.The catalytic converter, which oxidizes CO and hydrocarbons to CO 2 , probably reset the clumped signatures at relatively lower temperature.During combustion water vapor is also released.We observed that the exhaust gas contained a large amount of water vapor, part of which was condensed on the exhaust pipe and the front part of the magnesium perchlorate column.Partial equilibration with the stream of the exhaust gas and water inside catalytic converter and the exhaust pipe was the likely cause for higher 47 values than that expected.This was also supported by the higher δ 18 O values than atmospheric O 2 , the source of O 2 for water and CO 2 here.Normally isotopes in CO 2 do not exchange with water vapor, but exchange may take place at higher temperature in presence of catalyst.Inside the cat-alytic converter, exchange could take place on the surface of the catalyst at elevated temperatures of 200-400 • C (Farrauto and Heck, 1999; Kašpar et al., 2003;Klingstedt et al., 2006).Affek and Eiler (2007) also observed elevated 47 values for car exhaust and estimated a temperature of CO 2 production to be ∼ 200 • C. The temperature estimated here (283 • C) is significantly higher than that observed by Affek and Eiler (2007).Difference could be due to different car models and the variations in the temperatures of the catalytic converters from car to car.

Marine and coastal air CO 2
Carbon Keeling plot for marine and coastal air CO 2 gave an intercept of −13.61 ± 1.14 ‰ (Fig. 5a), the source signature.The South China Sea is net source of CO 2 to the atmosphere (Zhai et al., 2005).The CO 2 released over ocean is mainly originated from the remineralization of organic matter in the deeper ocean (Francois et al., 1993;Goericke and Fry, 1994).The δ 13 C value of such organic matter ranges between −20 and −30 ‰ in the tropical to subtropical oceans, the intercept observed here (−13.6 ‰) is much higher than this range, though the associated uncertainty is high due to a small span of isotopic values of the samples.A possibility is that the remineralized CO 2 gets equilibrated with the dissolved inorganic carbon before releasing to the atmosphere.Again, a complete equilibration of the CO 2 with the dissolved inorwww.biogeosciences.net/13/5297/2016/Biogeosciences, 13, 5297-5314, 2016 ganic carbon would lead to a δ 13 C value of released CO 2 to be −9 to −10 ‰ (Mook, 1986;Boutton, 1991;Zhang et al., 1995;Affek and Yakir, 2014); the observed value of the intercept (−13.6 ‰) was significantly less than this.Therefore, we infer that the CO 2 produced in the deeper ocean is partially equilibrated with the dissolved inorganic carbon before releasing to the atmosphere.The δ 18 O values of the surface sea water in the South China Sea region in summer (July-September) and winter (December-February) were about −1.7 and −0.6 ‰ (Ye et al., 2014).The sea surface temperatures in the summer and winter are about 28 and 24 • C, and the equilibrated δ 18 O values of the atmospheric CO 2 should be 38.9 and 40.7 ‰, respectively, assuming fractionation factors at the respective temperatures (Brenninkmeijer et al., 1983).Our observed values lie in the range of 40.4-41.0‰ (Table 3), consistent with the isotopic equilibrium values with the surface water.Therefore, we conclude that oxygen isotopes in near-surface air CO 2 over ocean are close to the isotopic equilibrium with the surface sea water.This conclusion was further supported by the observed 47 values which were found to be close to thermodynamic equilibrium with the underlying sea surface water at the sea surface temperature (Fig. 5b).This is due to the same water-CO 2 exchange time for the two species (Affek, 2013;Clog et al., 2015).Comparing this observation with the greenhouse data above, we conclude that δ 18 O and 47 behave similarly when equilibrium is achieved by simple water-CO 2 exchange but respond differently when photosynthesis is the main governing factor.Though carbonic anhydrase is also present in the surface ocean and marine phytoplankton does photosynthesize, δ 18 O and 47 in air CO 2 over the ocean show the values at thermodynamic equilibrium unlike the case of greenhouse CO 2 .The degree of deviation from thermodynamic equilibrium probably increases with the increase in photosynthetic activity.Normally photosynthesis by oceanic plants is much less compared to their terrestrial counterparts; the deviation from thermodynamic equilibrium by the oceanic photosynthesis, if present, is probably not detectable with the present measurement precision.Compared to δ 18 O, 47 is process sensitive and is not affected by the isotopic composition of substrate water.Given that the surface air temperature is better measured, we believe the clumped isotopes potentially provide good tracers for global carbon flux study involving CO 2 , complementing the commonly used species like [CO 2 ], δ 13 C and δ 18 O.
In the coastal stations, 47 values were similar to the thermodynamic equilibrium with the sea surface water at the temperature of ∼ 27 • C (Fig. 5b).The recoded air temperature during the sampling period over the coasts varied between 14 and 24 • C and was not reflected in the 47 values.We note that the samples were collected from two open spaces in the coasts where strong north and northeasterly winds overwhelmed, carrying air masses from the oceans towards the sampling locations (see Table S3 in Supplement).Therefore, we expect the major contribution was marine air with little influence from local processes, which could occasionally cause deviation from the thermodynamic equilibrium values.

Urban and suburban air CO 2
A significant fraction of anthropogenic CO 2 was present in the air CO 2 over the urban site, indicated by the [CO 2 ] as well as isotopic compositions including 47 .Anthropogenic contribution can be estimated following two-component mixing: δ = f anth × δ anth + (1 − f anth ) × δ bgd , where δ can be δ 13 C, δ 18 O or 47 , f is the corresponding weighting factor, and subscripts "anth" and "bgd" refer to anthropogenic and background, respectively.We take the "anthropogenic" end member as the isotopic compositions of the car exhaust values (Table 2) and "background" end member as values observed over the ocean (for δ 13 C and δ 18 O, Table 3) and thermodynamic equilibrium value at the mean ambient temperature of ∼ 20 • C (0.95 ‰ for 47 ) at the site during sampling.Assuming that the excess in [CO 2 ] above the background was originated from vehicular emission, the values of δ 13 C, δ 18 O and 47 in the urban site obtained using the mixing equation were −12.26, 37.68 and 0.809 ‰, respectively, which were similar to those observed (Table 4).47 is not a conserved quantity and a linear mixing is not valid when the δ 13 C and δ 18 O of the components are widely different (Affek and Eiler, 2006;Laskar et al., 2016a).In the present case, the isotopic compositions of the two components were not drastically different and fraction of anthropogenic CO 2 was much less (< 1/4) than the background CO 2 , and hence the error due to linear approximation was small (comparable to the uncertainty of measurement).Anthropogenic CO 2 can also be quantified using radiocarbon ( 14 C) as fossil fuels are highly depleted in 14 C (Miller et al., 2012); however, it cannot distinguish difference between CO 2 from two sources with modern carbon.
No systematic diurnal or temporal trend was observed in the 47 values in the suburban CO 2 during the sampling period (Fig. 7b).However, a weak trend was seen in δ 13 C and δ 18 O (not shown) in response to the seasonal variation of the carbon assimilation and oxygen isotopes in the rainwater (Peng et al., 2010;Laskar et al., 2014).This furthermore demonstrates that 47 behaves differently from [CO 2 ], δ 13 C and δ 18 O.Almost all measured 47 values were lower than that expected at the ambient temperature except 2 days: 9 November 2013 and 3 February 2014.δ 13 C values were also slightly lower than the background values.The reduced values of 47 could be due to contribution of CO 2 from combustion processes which produced CO 2 with low 47 values as discussed in Sect.4.2.We estimated the contribution of local anthropogenic emission in δ 13 C and 47 using the twocomponent mixing discussed above.The components were the background air CO 2 and car exhaust.The expected δ 13 C and 47 values of the mixture were −9.1 and 0.92 ‰, respectively.The observed 47 value was significantly differ-ent from that estimated from simple two component mixing, though it was not different for δ 13 C.After subtracting the local anthropogenic contribution from the observed 47 values, a difference of ∼ 0.026 ‰ between the observed and estimated remains for suburban station and it disappeared for urban station (see Table S4 in the Supplement).This was not obvious in δ 13 C probably due to larger variation.The lower 47 values in suburban station could possibly be due to kinetic effect during photosynthetic assimilation, partial contribution of marine air or a combination of them.It could also be due to underestimation of the anthropogenic CO 2 at the sampling spot.The regional background [CO 2 ] here could be lower than that assumed and the actual anthropogenic fraction of CO 2 could be higher.The marine air in the vicinity of Taiwan, which was at thermodynamic equilibrium with the surface sea water as discussed earlier, might have contributed partially to the air CO 2 at the sampling site.Varying contribution of marine air could explain the lower 47 values to some extent.The most plausible cause for observed deviation in the 47 values that cannot be accounted for by anthropogenic and marine alterations was photosynthesis, as discussed earlier for greenhouse CO 2 .This is not unreasonable, as the Academia Sinica Campus is surrounded by thick greeneries.
On 9 November 2013 and 3 February 2014, the 47 values were close to that expected at thermodynamic equilibrium (Fig. 7b).The 47 values on 9 November were not very different from the values reported for the previous or next days.However, the calculated thermodynamic equilibrium values on that day were relatively low due to high ambient temperatures; air CO 2 probably did not get enough time to equilibrate.On 3 February 2014, the 47 values were higher and comparable to the thermodynamic equilibrium values expected at ambient temperatures.A likely explanation is that the air on that day was a mixture of two components at the sampling region.A relatively strong wind from the southern land (Table S3 in the Supplement) contributed the air CO 2 and the higher 47 values were probably due to mixing of the local air with that transported from the south of Taipei.

Grassland, forest and high mountain air CO 2
In the grassland station in Taipei city, the Keeling plot intercept for δ 13 C (−17.0 ± 1.0 ‰) (Fig. 5d) indicated some sources of CO 2 with higher δ 13 C values compared to the most expected sources, namely C 3 vegetation and vehicle emission with a δ 13 C value of ∼ −27 ‰.Though the sampling station was located in an urban region, the sampling spot was at least ∼ 150 m away from traffic streets, such as Keelung road, along with ∼ 60 m wide, ∼ 10 m high C 3 trees in between.As a result, anthropogenic signals were not very prominent.The samples were collected just above the surface of the grasses.Tropical warm grasses are mainly C 4 type with δ 13 C in the range of −9 to −19 ‰ and a global average of −13 ‰ (Deines, 1980).We measured δ 13 C values of a few grass samples and found values in the range of −15 to −17 ‰.The soil and grass respired CO 2 with higher δ 13 C contributed significantly to the near-surface CO 2 , resulting in a higher value of intercept (−17 ‰).The concentration was observed to be less than the background level sometimes, probably due to strong CO 2 uptake by plants.The temperature gradually decreased from 26 to 20 • C during the consecutive 3 days and clumped isotope followed similar trend, reflecting the influence of temperature on CO 2 47 and rapid equilibration with the leaf and surface waters.One low value observed on the second day was probably due to plumes of vehicle exhaust, also supported by the elevated level of [CO 2 ] and depletion in δ 13 C and δ 18 O (Table 4).Effect of photosynthesis on the CO 2 was also expected specifically due the collection of samples at the grass level.However, in an open system, it is difficult to assess this with limited data points.
An elevated CO 2 concentration and low δ 13 C and δ 18 O values indicated significant contribution of respiration and/or anthropogenic CO 2 in the forest station (Table 4) near the Academia Sinica Campus.Though the samples were collected at 10:00-11:00 under bright sunlight, the vegetation was so dense that little sunlight reached the ground.Probably photosynthetic activity was not very strong at the ground level in the morning hours and the dominant process was respiration.Also, poor circulation of air due to presence of high heels on the three sides of the sampling spot made the site nearly isolated from the surroundings.As a result the 47 values were observed to be similar to the thermodynamic equilibrium expected at the ambient temperatures except on 11 August 2015 (Fig. 7f).This also supports our hypothesis, made in the case of greenhouse CO 2 , that respired CO 2 is in close thermodynamic equilibrium with the substrate water.On 11 August 2015 a significantly higher 47 value was observed.The higher value was likely due to the influence of the super Typhoon "Soudelor" which passed over Taipei during 8-10 August 2015, causing a decrease in temperature by 3-4 • C and air masses mixing in a larger spatial scale.
For high mountain CO 2 , the observed 47 values (Table 4) were lower than that expected at ∼ 10 • C, the ambient temperature at the top of the mountain site during sampling.The 47 values were similar to that observed in the plain and over the ocean.We note that during the sampling period, the site was affected significantly by winter monsoons.HYSPLIT 24 h back trajectory showed marine origin of air (not shown) during the sampling time.The air CO 2 on the mountain probably did not get sufficient time to isotopically equilibrate with the local surface and leaf water but showed the signature of the marine CO 2 .
The deviations in 47 from the thermodynamic equilibrium values in different atmospheric environments and processes are summarized in Fig. 9.It is obvious that the urban and suburban CO 2 deviate the most towards lower 47 values, mainly contributed by CO 2 originated from high-temperature combustions, i.e., vehicular emissions.The respired CO 2 are in close thermodynamic equilibrium at the www.biogeosciences.net/13/5297/2016/Biogeosciences, 13, 5297-5314, 2016 ambient temperature.In contrast, CO 2 affected by strong photosynthesis shows significant increase in the 47 values compared to the thermodynamic equilibrium values.

Summary
We presented a compilation of 47 analyses for car exhaust, greenhouse and air CO 2 over a wide variety of interactions in tropical and subtropical regions including marine, coastal, urban, suburban, forest and high mountain environments.Nearsurface marine air CO 2 is in close thermodynamic equilibrium with the underlying surface water at the sea surface temperature.Car exhaust, urban, suburban and greenhouse air CO 2 significantly deviate from the thermodynamic equilibrium values.While respired CO 2 is in thermodynamic equilibrium with leaf and soil surface waters, photosynthesis significantly deviates the 47 values from the thermodynamic equilibrium or more precisely increases the 47 values probably due to kinetic effect associated with the diffusion of CO 2 out of leaf stomata.The 47 values in urban and suburban air CO 2 are lower than that expected at thermodynamic equilibrium at the ambient temperature.The deviation is mainly due to contributions from fossil fuel emissions and to some extent due to photosynthesis, especially in regions with dense vegetation.
We showed that 47 can serve as an independent tracer for studying photosynthesis.Though the deviation from equilibrium during photosynthesis is also observed in oxygen clumped isotopes, CO 2 and O 2 are affected and produced from different processes and sources; the former is affected seriously by water (water-CO 2 isotopic exchange) while the latter is derived from water.We believe the analyses of the clumped isotopes for both CO 2 and O 2 are of great importance in the atmospheric carbon cycling study, providing a new angle for tackling the chemistry chain in photosynthesis.More systematic study in controlled environments including leaf level experiments will help to better understand the role of photosynthesis on 47 .

Data availability
All the data used in the manuscript are also presented in the form of tables.
The Supplement related to this article is available online at doi:10.5194/bg-13-5297-2016-supplement.

Figure 1 .
Figure 1.Left panel: map of Taiwan and South China Sea with the locations of marine air sampling stations (a-e).The coastal stations 1 and 2 are Fugui Cape and Keelung and 3 is the high mountain station Hehuan (∼ 3.2 km a.s.l.).Right panel: part of Taipei city with sampling stations Roosevelt Road (RR), grassland in the National Taiwan University (NTU) Campus, suburban site inside the campus of Academia Sinica (AS) and forest site.

Figure 2 .
Figure 2. Diurnal variation of (a) concentration, (b) δ 13 C and (c) δ 18 O of CO 2 sampled in the greenhouse.Keeling plots for (d) δ 13 C and (e) δ 18 O and (f) scatter plot of δ 13 C and δ 18 O to show their covariance.

Figure 3 .
Figure 3.Diurnal variation of the 47 and δ 18 O values in the greenhouse for samples collected on 4 days of 2015: (a) 28 July, (b) 31 July, (c) 4 August and (d) 12 October.The first 3 days (a-c) were bright sunny days and the last was (d) a cloudy day with covered rooftop (see texts for details).The error bars are 1 standard error associated with the measurements.

Figure 4 .
Figure 4. Correlation between the observed and thermodynamic equilibrium 47 values for greenhouse CO 2 samples collected when (a) photosynthesis was weak and respiration was strong and (b) photosynthesis was strong and respiration was weak.

Figure 5 .
Figure 5. (a) Carbon Keeling plot for air CO 2 collected over South China Sea (blue symbol) and coastal stations (pink symbol) (Keelung and Fugui Cape).(b) 47 values observed over the South China Sea and coastal stations.The error bars are the 1 standard error associated with the measurements.Lines show 47 values for the CO 2 in thermodynamic equilibrium at ambient temperatures.

Figure 6 .
Figure 6.Carbon Keeling plots for air CO 2 collected over (a) suburban Academia Sinica Campus and (b) grassland at National Taiwan University Campus.

Figure 7 .
Figure 7. 47 values in the near-surface atmospheric CO 2 from (a) urban site near Roosevelt Road on 30 December 2015, (b) suburban station (Academia Sinica Campus), (c) grassland in the National Taiwan University Campus and (d) forest site near the Academia Sinica Campus.The error bars are the 1 standard error associated with the measurements.Lines show 47 values for the CO 2 at thermodynamic equilibrium at ambient temperatures.

Figure 8 .
Figure 8.(a) CO 2 concentration inside greenhouse on 31 August 2015: observed concentration (star) and decrease in concentration by photosynthesis after subtracting the respiration (solid circle) are also shown.Comparison of observed (b) δ 13 C, (c) δ 18 O and (d) 47 values with that modeled using discrimination factors of −16.5, −12.0 and 0.065 ‰ for δ 13 C, δ 18 O and 47 , respectively.

Figure 9 .
Figure 9.A summary of 47 values in near-surface air CO 2 obtained at different environments and compared with the thermodynamic equilibrium values.Combustion temperature for car exhaust is assumed to be 800 • C (minimum value).Greenhouse CO 2 are divided into two categories: photosynthesis dominated (green open circle) and respiration dominated (green open triangle).

Table 1 .
Diurnal variation of δ 13 C and δ 18 O and clumped isotopes ( 47 ) for greenhouse CO 2 .Temperatures estimated using 47 values and actual air temperatures inside the greenhouse at the time of sampling are also presented.

Table 2 .
Stable carbon and oxygen isotopic composition and clumped isotopes ( 47 ) for car exhaust CO 2 .Temperatures estimated using 47 values and lowest possible combustion temperatures are given.

Table 3 .
Stable isotopic composition including 47 for atmospheric CO 2 collected over South China Sea and two coastal stations (see Fig.1for sampling locations).Temperatures estimated using 47 values and the sea surface temperatures at the time of samplings are also presented.

Table 4 .
Stable isotopic composition including clumped isotopes ( 47 ) for air CO 2 collected in urban and suburban stations, grassland, forest and high mountain environments.Temperatures estimated using 47 values and air temperatures are also presented.