Geophysical Research Letters

Riverine inorganic carbon flux and rate of biological uptake in the Mississippi River plume



[1] Inorganic carbon parameters were studied for the first time in the Mississippi River estuary and plume. Area-integrated biological uptake rates (1.5–3 gC m−2 d−1) derived from riverine total dissolved inorganic carbon (DIC) and total alkalinity (TAlk) fluxes as well as that based on the consumptions of DIC and TAlk in the plume are comparable to earlier results of 14C incubation in individual water samples. This rate is among the highest in the estuaries and plumes of the world's largest rivers. Fluvial DIC flux (13.5 × 1012 gC yr−1) indicates a 16% increase in the inorganic carbon flux over the earlier estimate and perhaps an increase in the weathering rate in the Mississippi River basin over the past four decades.

1. Introduction

[2] The Mississippi River ranks seventh in freshwater discharge among world's major rivers. Along with freshwater, it supplies a large amount of dissolved and suspended materials to the Louisiana-Texas shelf and it sustains high primary production there. High productivity and stratification also introduce a large area of hypoxia along the coast, which is aggravated by anthropogenic nutrient inputs. As a result, this area has been a focus of study by many investigators using various approaches including O2 and nutrient dynamics, microbial processes and primary production [Lohrenz et al., 1997, 1999 references therein]. However, no inorganic carbon approach has been taken to the study of the biological processes, i.e., photosynthesis and respiration, where CO2 is the primary participant. Here, I take a new approach that estimates an area-integrated biological uptake rate in the Mississippi River estuary and plume based on riverine inorganic carbon fluxes or the inorganic carbon depletion in the plume. Riverine inorganic carbon flux and weathering rate in the drainage basin will also be discussed.

2. Site and Methods

[3] Cruises were taken aboard R/V Pelican in August and September 1998. Our research area covers the lower Mississippi River (the estuary) and the area adjacent to the birdfoot delta outside the Southwest Pass (Figure 1). The latter is correspondent to the central shelf areas in Lohrenz et al. [1997]. The sampling sites are also generally in the plume identified by the Lagrangian drifter in Hitchcock et al. [1997]. The Mississippi River plume is highly stratified with a layer of fresher water in the top (∼5 m).

Figure 1.

Area map with sampling sites and salinities.

[4] Surface water samples were taken with a Niskin bottle. For total dissolved inorganic carbon (DIC), total alkalinity (TAlk) and Ca2+ analyses, water samples were stored in 250-mL glass bottles preserved with 20 μL saturated HgCl2. DIC was measured shortly after the cruise using a DIC analyzer; following acidification of 0.5 mL of sample and quantification of the released CO2 in an infrared detector (Li-Cor 6252). The method has a precision of 0.1% (modified from Cai and Wang, 1998). TAlk was determined by Gran titration on 40 mL of sample with a Kloehn digital syringe pump. It has a precision of 0.1%. Ca2+ was measured on 6–8 grams of sample by EGTA titration [Tsunogai et al., 1968] and the uncertainty of the method is 0.1%. Surface water samples were also collected without air contact in a 60-mL syringe and then injected into a flow-through cell for pH measurements with an Orion Ross combination glass electrode that was calibrated against three NBS standards. pCO2 was calculated from measured pH and DIC based on constants suggested for estuarine waters in Cai and Wang [1998].

3. Results and Discussion

3.1. River End-Member Signals

[5] Both DIC and TAlk were very high (2.2 to 2.9 mM) in the near-zero salinity zone of the Mississippi River (Figure 2). Such high DIC and TAlk values were not seen in any other large rivers except the Huanghe (Yellow River) (TAlk ∼3.0 mM, Sun et al., 1993). The Amazon River has DIC and TAlk values around 0.6 mM [Ternon et al., 2000]. The Pearl River (in South China) has DIC and TAlk values around 1.0 mM [Cai unpub.]. The high inorganic carbon content of the Mississippi River and the Huanghe as compared to the subtropical Pearl River and the tropical Amazon River reflects different weathering intensities at various climate zones.

Figure 2.

DIC, TAlk and Ca2+ systematic in the Mississippi River estuarine and plume collected on September 25, 1998. Also plotted are anchor station data on September 24 and on August 8, 1998 at a near zero salinity location. Surface water DIC and TAlk values of the Atlantic Ocean are taken from Millero [1996]. Mixing lines are calculated based on values at the two end-members (S = 0.25 as the river end-member). For Ca2+ the mixing line is very closed to a linear regression line that is shown in the figure with an R2 = 0.999 ([Ca2+]/μM = 261.3 * S + 1138.7). The dashed lines are drawn only to assist the view of biological uptake.

[6] The Mississippi River end-member near New Orleans had a salinity of 0.02 in August 1998 and 0.2 in September 1998. The higher salinity in September is most likely a result of diminished freshwater discharge. Daily average discharge decreased from 15.9 × 103 m3 s−1 in August to 5.9 × 103 m3 s−1 in the 10-day period before the sampling in September 1998. DIC (or TAlk) increased from 2155 (or 2115) μmol kg−1 to 2920 (or 2870) μmol kg−1 during this period of decreased discharge. The observed DIC and TAlk changes are most likely a result of the dilution of the weathering signal by the river discharge. Riverine DIC fluxes calculated based on August 1998 (near average river discharge) and September 1998 (at low flow) are 2.96 × 109 molC d−1 and 1.49 × 109 mol C d−1 respectively. It appeared that the dilution of DIC and TAlk signals (20%) was far from enough to compensate for the increase in the discharge rate (2.7X). The daily average discharge rate of the river at Tarbert Landing, LA between 1930 and 2000 is 13.7 × 103 m3 s−1 and is 17.1 × 103 m3 s−1 for 1998.

3.2. Biological Uptake in the Mixing Zone

[7] Beyond salinity 0.3 psu, both DIC and TAlk decreased almost linearly until at about 29–30 psu (Figure 2). From the site map and the sample time (within 30 min), it was clear that water with a salinity of 3.8 and waters with salinities of 24–30 psu were in the vicinity of each other and were right outside the river mouth where a strong salinity gradient exists. Thus, biological removal might not be discernable due to the fast mixing. However, the DIC values in S = 24 to 35 were clearly lower than that of seawater (Figure 2). Furthermore, the highest salinity end member was also clearly lower in DIC but not much lower in TAlk when compared to North Atlantic Ocean surface water (Figure 2).

[8] In the salinity range of 0.3 to 35.5, one may treat the data as a two end-member mixing situation with a maximum removal occurring at salinities 29–30; or one may argue that this is a mixing of three end-members (Figure 2). However, there is not a viable third end-member there that has a larger volume (> river flow), a high salinity (>30) and a low DIC (<1900 μM). The Mississippi plume inorganic carbon system is essentially a mixture of freshwater and oceanic water and can be approximated with a two end-member mixing model. The linear correlation between [Ca2+] and salinity also supports this two end-member approach (Figure 2).

[9] In the following paragraphs, the rate of biological uptake in the Mississippi River plume will be estimated with two approaches. The first approach is based on the mixing curve and a standard estuarine mixing model. Applicability and caution of using such a model were discussed in earlier publications [Officer, 1979]. At a certain salinity, a tangent line can be drawn on the DIC∼S curve and can be extended to the zero salinity. The DIC thus obtained is called the effective concentration. A product of the river discharge rate (Q) and this effective concentration (C*) represents the flux of a dissolved material across that iso-salinity surface (Flux = QC*). Since before salinity 29, the DIC ∼ S plot is almost a straight line with an effective concentration of about 2920 μmol kg−1, the DIC flux can be estimated as 1.49 × 109 molC d−1, which is the same as the fluvial flux. The slope of the DIC∼S curve changes sign at salinity around 30. The effective concentration for the segment between salinities 29 and 35 is about 1380 μmol kg−1 and the flux is 7.03 × 108 molC d−1. The amount of DIC consumed in the plume is roughly the difference of these two fluxes, 7.87 × 108 molC d−1. Here we assume that the DIC and TAlk values measured in the main plume (the Central area outside the Southwest Pass) is applicable to all Mississippi River discharge. Dividing this number by the mixing area of the plume, a biological uptake rate in the early stage of the Mississippi plume can be estimated. The East and Central areas of the Mississippi plume were estimated as 6600 km2 [Lohrenz et al., 1997]. For a particularly low flow period, the area of the plume might be shrunk. If a 25 to 50% reduction of the area is applicable, the biological uptake rate is derived as 160–240 mmol m−2 d−1 (or 1.9–2.8 gC m−2 d−1). Similarly, the TAlk uptake rate is estimated as 60–90 mmol m−2 d−1 (or 0.73–1.1 gC m−2 d−1). Since DIC is up taken into both organic carbon and CaCO3, the amount of DIC synthesis into organic carbon shall be corrected for half of the alkalinity uptake and is thus 130–190 mmol m−2 d−1 (or 1.5–2.3 gC m−2 d−1).

[10] The second approach is based on the amount of DIC depletion in the plume and the plume travel time. At S = 30, the DIC depression from the conservative mixing line is 226 μmol kg−1 and the TAlk depression is 114 μmol kg−1. Therefore, DIC loss due to organic carbon synthesis is 226–114/2 ≈ 180 μmol kg−1. Such DIC and TAlk consumptions are comparable to those observed in the Amazon plume (200 and 172 μmol kg−1 respectively, Ternon et al., 2000) except that the latter has a much longer plume travel time. The residence time of the Mississippi plume varies, but a time of 1–2 days has been estimated [Lohrenz et al., 1999, 1997; Hitchcock et al., 1997]. A 2.5-day residence time is used here since September 1998 is a low flow period. The thickness of the plume is also related to the river discharge rate. An average of 4.7 m was estimated for March of 1993 when monthly river discharge was 24.4 m3 s−1 [Lohrenz et al., 1997]. In a low flow time as September 1998, the plume thickness might be shrunk significantly. Again, if a 25–50% reduction in the plume thickness was applicable, then a biological uptake rate of 170–250 mmol m−2 d−1 or 2.0–3.0 gC m−2 d−1 is estimated (i.e., 180 mmol m−3 * 4.7 m/2.5 d). This estimate may have an uncertainty of 35% if the plume thickness and its travel time each has a 25% uncertainty. If the plume receives significant input from the subsurface water (i.e., a third end member) then a longer time scale must be assumed, and therefore a lower uptake rate would be estimated. Considering the uncertainties involved in the physical parameters used here, the above two estimations are about the same.

[11] The area-integrated biological uptake rates estimated from riverine DIC flux or from the DIC depletion in the plume are consistent with individual water sample-based primary production rates using 14C incubation technique. For the April 1988 cruise, primary production rates were mostly within 0.5–3 gC m−2 d−1 [Lohrenz et al., 1997], and higher values (mostly 0.5–8 gC m−2 d−1) in later years were reported [Lohrenz et al., 1997, 1999]. It should be noted that our estimate represents a net ecosystem production (NEP) of the surface water while the incubation method represents somewhat between net primary production (NPP) and gross primary production (GPP) [Bender et al., 1987]. NEP is much smaller than GPP if the synthesized organic carbon is respired mostly within the system. The Mississippi plume, however, has a high NEP as its synthesized organic matter rains out of the plume or does not respire within the time scale of mixing.

[12] It also appears that the biological uptake rate estimated based on DIC and Alk depressions in the Mississippi plume is higher than that in the Amazon plume (1.2–1.6 gC m−2 d−1, Ternon et al., 2000). This might be a result of the higher nutrient concentrations in the Mississippi River.

3.3. Supporting Information From pH and pCO2

[13] The fluvial end-member had a relatively high pH of 7.98 and a high pCO2 of around 2000 μatm (Figure 3). The high pH was consistent with the high TAlk of the river and reflected the weathering characteristics in its drainage basin. An elevated pCO2 was observed in many rivers and estuaries, and was interpreted to be a result of microbial respiration and/or groundwater input [Kempe et al., 1982; Frankignoulle et al., 1996, Cai and Wang, 1998]. Although with a high pH and a high TAlk, the fluvial end-member had a very low [CO32−] of 18 μmol kg−1 and a high [CO2] of 58 μmol kg−1.

Figure 3.

pH and pCO2 values in the Mississippi estuarine and plume waters. Dotted line indicates pCO2 = 370 μatm.

[14] The pCO2 value dropped quickly to about 1000 μatm as pH increased before reaching the river mouth. During the initial mixing stage in the plume, pH increased to as high as 8.4 while pCO2 decreased to below the atmospheric level of ∼370 μatm. Such high pH and low pCO2 are consistent with a high photosynthetic activity and low DIC value in this section of the plume. At the later stage of the plume (S ≥ 35), pH decreased and pCO2 was at equilibrium with the atmosphere. This is consistent with a higher DIC and was mostly a result of organic matter decomposition in the plume.

4. Synthesis

4.1. Is There an Observable Biological Uptake?

[15] Lohrenz et al. [1997, 1999] have studied the primary production and nutrient dynamics of the Mississippi plume in great detail. The primary production rates measured were among the highest in large river estuarine and shelf waters. In this paper, we have presented new evidence of significant biological uptake in the Mississippi plume from an inorganic carbon and a system level perspective.

[16] Using Lagrangian drifters as a tracer, Hitchcock et al. [1997] found that nutrient concentrations followed conservative mixing behavior within the plume and that removal occurred mainly in the plume edge. These different observations really are not in contradiction. First, some of the data reported by Lohrenz et al. [1999] were collected from the plume edge. Second, there was a huge difference in the river discharge rates in the cruises that were carried out by these two groups (Table 1). The two cruises in 1993 with no observable NO3 removal in the plume were on a particularly high river discharge year. Biological removal was observed in most of the low flow cruises (Table 1). It appears that whether or not the uptake can be observed depends on the discharge rate.

Table 1. A Comparison of River Discharge Rate and Assessment of Biological Removal
Cruise timeDischarge (m3 s−1)*Removal within the plumeRefs
  • *

    Daily average starting 10 days before the cruise and ending on the last day of the cruise.

  • **

    Release at S < 28, removal S > 28. This is also somewhat true for the other “not clear” cases. 1. Lohrentz et al. [1999]. 2. Hitchcock et al. [1997]. 3. This work.

Apr. 16–24, 8820.9not clear1
Sept. 20–27, 899.7**1
July 17–Aug. 10, 9013.9clear removal1
Mar. 4–17, 9126.3not clear1
Sept. 12–18, 916.3clear removal1
May 5–18, 9214.6clear removal1
May 21–22, 9333.0no removal2
Aug. 23–24, 9323.2no removal2
Sept. 25, 985.9DIC removal3

4.2. Weathering Rate and Riverine Flux

[17] It was shown that nitrate concentration correlated positively with the river discharge rate in the Mississippi River [Lohrenz et al., 1997, 1999]. Higher discharge appears to release more nutrients from the soil in the drainage basins that have been heavily fertilized. Interestingly, DIC was negatively correlated to the river discharge in the available dataset (1997–2000, Figure 4). HCO3 is the major anion resulting from the weathering process. The negative and nearly linear correlations suggest that weathering rates in the drainage basin are more stable than the nitrate release rate, and the river HCO3 concentration is somewhat of a dilution of the weathering signal in the drainage basin. The correlation coefficient is, however, not very high (R2 = 0.6, Figure 4), indicating that other factors also affect the weathering rate. Riverine HCO3 flux is, however, highly and positively correlated to the discharge rate (Flux = −7E–17Q3 + 8E–08Q2 + 1919.1Q + 3E+11, R2 = 0.93).

Figure 4.

Correlation between the bicarbonate concentration and the discharge rate at St. Francisville, LA, which is upstream of New Orleans, LA. The bicarbonate concentration (in μM) can be correlated to the discharge (Q in m3 s−1) with a linear or a third order polynomial equation: [HCO3] = −0.0304Q + 2648.7 (R2 = 0.56) or −5.96E-11Q3 + 4.32E-06 Q2 − 1.18E-01Q + 3.11E+03, (R2 = 0.60). Data are from the USGS database “Water-Quality Data for the Nation” ( A few data points (superimposing of o and x) were not available in the USGS database and were added in by scaling the DIC value of the previous date to TAlk data of the two dates (i.e., DICb = DICa × TAlkb/TAlka).

[18] The USGS HCO3 data agree well with our measurements at the fluvial end member. The large HCO3 increase from July 30 (2444 μM) to September 28 (2902 μM), 1998 was recorded in the USGS data and was consisting with the decrease in the flow rate and with our data. The average HCO3 value and flux of the Mississippi River between 1996 and 2000 (Figure 4) are 2133 μM and 2.92 × 109 molC d−1 (12.5 × 1012 gC yr−1). TAlk data (also from USGS; not shown here) in the same period have a similar average value and flux of 2159 μM and 3.04 × 109 mol d−1. The difference indicates an average [CO32−] of only 13 μM. From these values and carbonate equilibrium calculation, an average [CO2] of 44.5 μM and an average DIC = 2190.5 μM are derived. These results are very similar to our August 1998 results. The average DIC (=[CO2] + [HCO3] + [CO32−]) flux thus estimated is 3.08 × 109 molC d−1 (or 13.5 × 1012 gC yr−1). This value is 16% higher than the earlier estimate of 11.6 × 1012 gC yr−1 [Kempe, 1982 and Degens et al., 1991], and indicates a possible increase in weathering rate in the Mississippi River drainage basin in recent decades. TAlk data between 1964 and 2000 indicate the same trend with increasing average fluxes of 2.27, 2.52, 2.73 and 2.98 × 109 molC d−1 for the periods of 1964–1973, 1974–1983, 1984–1993 and 1994–2000. Changes in land use patterns may be a cause of increased weathering rate. It is also speculated here, that in addition to climate zone, land use pattern also contributes to the exceptionally high DIC and TAlk values in the Mississippi River and the Huanghe given that both river drainage basins are in populated regions.


[19] The author thanks B. McKee for his invitation to the cruise, T. Bianchi for the site map, and D. Walters, Data Management Supervisor, Louisiana District, USGS, J. A. Moody and A. Shiller for assistance in finding the USGS database. L. Pomeroy, E. Turner and two reviewers had provided valuable inputs. Y. Wang helped with data collection.