The Changjiang drainage basin is situated within 24°30'–35°45'N and 90°33'–122°25'E, draining almost one fifth of the total area of China. The major tributaries in this region include Yalongjiang, Minjiang, Tuojiang, Jialingjiang, Wujiang, Hanshui, Xiangjiang, Yuanjiang, and Ganjiang (Fig. 1). Xiangjiang and Yuanjiang join into the main channel via Lake Dongting and Ganjiang via Lake Poyang. Lake Poyang and Dongting are the two largest freshwater lakes in China 27.

Datong Hydrographic Station (DHS; 117°37'E, 30°46'N as shown in Fig. 1), located 625 km inland from the river mouth on the mainstream, is free from tide influence. There is no major sewage discharge from the industrial and domestic sources nearby. DHS gauges the drainage area of 1.71 × 10⁶ km², i.e., 95% of the Changjiang watersheds. Therefore, data from DHS have extensively been used to evaluate the variation of water quality of Changjiang and to calculate the seaward flux by the world river in the literatures.

DHS was established in 1935. In 1956, the Ministry of Water Resources of China (MWRC) started to systematically monitor the hydrological parameters and water quality along Changjiang, as well as other rivers in China 28. Data were reported in the internal (unpublished) Hydrological Yearbooks. Monitoring data of water chemistry at DHS over the period 1962 – 1984 presented or used in this study was extracted from the unpublished Hydrological Yearbooks 29. Data over the period 1985–2000 was officially provided by DHS since data compilation in hydrological yearbooks has been discontinued after 1985. All monitoring data were processed by month, while the annual values were obtained by discharge weighted chemical parameters.

According to station, water was sampled twice per month at DHS. Chemical analyses of the water samples were performed in the laboratory under the authority of the Changjiang Water Resource Commission and following the methods described by Alekin et al. in 1973 30 and by the American Public Health Association in 1985 31. The MWRC also issued its own water quality analytical methods (SL78~94-1994). Specifically, raw water samples were siphoned to determine dissolve CO₂ concentration using base titration. This method is suitable for surface water analysis, and its standard deviation was reported as 0.105–9.81% when analyzing 17 different kind water samples with dissolved CO₂ concentration ranging from 2.73–2028 mg/L (SL80-1994). Dissolved water samples were decanted or filtrated through a 0.45-μm membrane from the raw water samples, and analyzed for HCO₃^- (by acid titration, with a standard deviation of 0.71%), and SiO₂ (by the Hetropoly blue method, with a standard deviation of 1.01%) concentration.

Water samples were filtered through filter papers before 1987 and 0.45-μm filter membranes after 1987 prior to nitrate determination. The filtrates were measured using spectrophotometry with a phenol disulfonic acid before 1987 and UV spectrophotometry after 1987. The phenol method is suitable for measuring concentration ranging from 0.02–2.0 mg/L with a standard deviation of 6.7%, while the UV method is suitable for the concentrations from 0.08–4.0 mg/L with a standard deviation of 1.1%. Both methods are the current Chinese standard for nitrate determination. Therefore, the change of methods had a negligible impact on the accuracy of the data, as has been discussed in detail by Yan et al 21.

Generally, no detailed quality assurance/quality control information is available on historical monitoring data from the annual hydrological reports. Chemical elements in Changjiang are measured at the mg/L scale and the analytical methods are routinely practiced in the national laboratories. Data reported in the yearbooks are therefore considered to be reliable. In fact, Chen et al. (2002) compared the data from another hydrological station on the Changjiang mainstream (Wuhanguan) with data monitored by GEMS/Water Program at the same station. It was found that two sets of data were in good agreement, indicating the reliability of data in the hydrological yearbooks. For detailed reliability analysis of the data in the hydrological yearbooks, see Chen et al., 2000, and Yan, et al., 2003 2127.

Fig. 2 illustrates the seasonal variations of water discharge at DHS in the lower reach of Changjiang, where the maximum monthly water discharge occurs in July and the minimum occurs in January. The magnitude of variation (five folds difference) of water discharge in July and January draws attention to the substantial water input from precipitation in the summer flood season. In addition, as shown in Fig. 2, the annual average of water discharge and minimum discharge were relatively stable, with a slight increase of maximum discharge in the period 1956–2000. It indicates a steady hydrographic feature of Changjiang in the past decades.

HCO₃^- is mainly derived from the weathering reaction of carbonate and silicate minerals with atmospheric CO₂, for which atmospheric contribution represents one half of the alkalinity in the case of carbonate weathering and all in the case of silicate weathering. Given the fact that carbonate rocks are well spread in the Changjiang drainage basin, the major element composition of Changjiang and its tributaries is dominated by the HCO₃^- anion. On average, HCO₃^- accounts for 64% (in mg/L units) of the TDS 27. From Fig. 3, it is clear that there was no clear temporal increase in the HCO₃^- concentrations over the period of 1960s–2000, whereas its inter-annual variation fluctuates between 1.2 and 2.4 mM, with an annual average of 1.7 mM. At the opposite of Changjiang, it was reported that for the Mississippi (the largest river in North America), the export of alkalinity has dramatically increased during the past half-century 32. The authors ascribed this increase to the change in land cover and heavier rainfall, which resulted in an increase in the rate of chemical weathering. Generally speaking, the rate of long-term chemical weathering involving CO₂ is mainly correlated with precipitation, stream flow, and temperature 33. With regard to Changjiang, however, such increase was not observed (Fig. 2, 3), indicating that chemical weathering rate in the Changjiang basin had been maintained a constant for the past decades.

To illustrate the seasonal variation, monthly change of the HCO₃^- concentrations in period of 1964–2000 are plotted in Fig. 4. It reveals that HCO₃^- concentrations was measured high in the winter, and low in April through May (Fig. 4a). Although the water discharge of Changjiang showed negative relationship with its HCO₃^- concentration (Fig. 4d), it reached a maximum in July (Fig. 2). This means that dilution was compensated by the effect of increasing dissolution of carbonates as a result of an intensification of soil erosion and elevated pCO₂ during flood season 27.

Data for the past decades (1964s–2000) shows that dissolved CO₂ concentration of Changjiang has declined dramatically with discharge weighted annual average being reduced from >100 μM in the 1960s to nearly 30 μM by the end of the 1990s (Fig. 3). A peak of dissolved CO₂ was observed around 1968. Compared with data after the 1970s, monthly concentrations average of dissolved CO₂ in the 1960s had a higher fluctuation (Fig. 4c). Collectively, water discharge had a less impact on changing the dissolved CO₂ concentration (Fig. 4d).

In general, dissolved CO₂ in water will reach equilibrium with the atmosphere CO₂, as described by the Henry's Law 34:

CO₂ (g) + aq = H₂CO₃*

K_H = [H₂CO₃*]/pCO₂ (M atm^-1)

Where, aq stands for the water phase; K_H is the Henrys Constant; pCO₂ is the partial pressure of CO₂.

H₂CO₃* is mainly composed of [CO₂] (i.e. dissolved CO₂) in the natural water phase. In this study, we calculated pCO₂ in Changjiang based on dissolved CO₂ concentrations according to Equations (1, 2). Results showed that the pCO₂ in the Changjiang riverine system is significantly higher than that of the atmosphere (350 ppm), and it appeared a clear trend of reduction since the 1960s (Fig. 5). In addition, few data in the 1990s have become comparable to the equilibrium concentration of atmospheric CO₂ (Fig. 5). In fact, it is often the case that CO₂ is over-saturated in rivers worldwide (Table 1). Compared with rivers listed in Table 1, pCO₂ in Changjiang was 1297 ± 901 ppm, lower than that of other large rivers like Amazon or the Mississippi.

CO₂ dissolved in river has sources mainly from the atmosphere CO₂, respired CO₂ and dissolution of carbonate 15. Because pCO₂ in Changjiang was clearly higher than that of the atmosphere, Changjiang should be a potential source for CO₂ emission to the atmosphere. The basin weathering (such as soil respiration, carbonate dissolution) generally provides a stable input of CO₂ to river water due to the constant chemical weathering rate. Consequently, in-channel respiration and photosynthesis, i.e. change in water quality, can be responsible for the decline of dissolved CO₂ in Changjiang.

It is generally accepted that CO₂ exchange between rivers and atmosphere is an important but still uncertain factor in the global carbon cycle 56. Recent studies estimated that, nearly 0.5 Gt C.yr^-1 escapes into the atmosphere from the Amazon basin through air-water CO₂ exchange, almost an order of magnitude higher than the organic carbon flux from Amazon to the ocean 13. Since rivers in the world are generally CO₂ over-saturated (Table 1), the gross emission of CO₂ from rivers into the atmosphere can be substantial, and it can balance off some important terrestrial carbon sink 35. On the contrary, the emission flux of CO₂ from Changjiang had remarkably decreased due to the downtrend of dissolved CO₂ concentration since the 1960s (Fig. 5). This can greatly affect the estimation of overall CO₂ exchange flux between Changjiang and the atmosphere, even at the global scale.