Carbon Dynamics on the Louisiana Continental Shelf and Cross-Shelf Feeding of Hypoxia

Abstract

Large-scale hypoxia regularly develops during the summer on the Louisiana continental shelf. Traditionally, hypoxia has been linked to the vast winter and spring nutrient inputs from the Mississippi River and its distributary, the Atchafalaya River. However, recent studies indicate that much of the shelf ecosystem is heterotrophic. We used data from five late July shelfwide cruises from 2006 to 2010 to examine carbon and oxygen production and identify net autotrophic areas of phytoplankton growth on the Louisiana shelf. During these summer times of moderate river flows, shelfwide pH and particulate organic carbon (POC) consistently showed strong signals for net autotrophy in low salinity (<25) waters near the river mouths. There was substantial POC removal via grazing and sedimentation in near-river regions, with 66–85 % of POC lost from surface waters in the low and mid-salinity ranges without producing strong respiration signals in surface waters. This POC removal in nearshore environments indicates highly efficient algal retention by the shelf ecosystem. Updated carbon export calculations for local estuaries and a preliminary shelfwide carbon budget agree with older concepts that offshore hypoxia is linked strongly to nutrient loading from the Mississippi River, but a new emphasis on cross-shelf dynamics emerged in this research. Cross-shelf transects indicated that river-influenced nearshore waters <15 m deep are strong sources of net carbon production, with currents and wave-induced resuspension likely transporting this POC offshore to fuel hypoxia in adjacent mid-shelf bottom waters.

Appendix

End-member values and AZE calculations associated with Fig. 3

The detailed end-member values used in constructing Fig. 3 are summarized in the following table and text.

Table 3 End-member values for river-ocean mixing. Values are means + SEM (n)

The detailed river end-member data used in Fig. 3 were averages of the July samples taken at Baton Rouge, with respective 2008 and 2009 means in millimoles per cubic meter + SEM (n) of 2,784 + 104 (4) and 2,697 + 91 (2) for DIC and 145 + 8 (4) and 135 + 6 (4) for nitrate + nitrite. The DIC deficit expected at 0 salinity if all nitrate + nitrite reacted was calculated assuming that phytoplankton growth was responsible for the deficit, with 85 % of DIC removal occurring to POC at the molar C/N Redfield ratio of 6.625 (Redfield et al. 1963) and the remaining 15 % (O’Reilly et al. 1987) of DIC forming DOC with a C/N ratio of 15 for mid-salinity summer river plume samples for this area (Table 2 in Pakulski et al. 2000; Table 4 in Benner and Opsahl 2001). DOC measurements made for samples collected during a July 2011 shelfwide cruise (data not shown) indicated a DOC excess of 100–200 mmol m⁻³, in agreement with expectations that DOC accounted for about 15 % of total DIC removal.

Similarly, the POC expected at 0 salinity if nitrate + nitrite reaction was complete assumed 85 % of DIC removal to POC occurring with the molar C/N Redfield ratio of 6.625. A value of 13 mmol m⁻³ representing estimated July Mississippi River POC (Fig. 2) was added to this value based on nitrate + nitrite consumption. Overall, calculations were done separately for 2008 and 2009 end-member values, then averaged to obtain values shown in Fig. 3. These averaged July river end-members in millimoles of carbon per cubic meter are 2,741 for DIC, 1,638 for DIC if all nitrate + nitrite reacts, and 801 for POC. The value of 801 mmol m⁻³ for POC after phytoplankton growth includes a baseline July estimate of 13 mmol m⁻³ for riverine POC obtained from low salinity samples collected during the shelfwide cruises (Fig. 2). The most important error associated with this end-member POC estimate of 801 mmol m⁻³ stems from the variability in nitrate + nitrite concentrations, with SEM values in 2008 and 2009 of 8 and 3 mmol m⁻³ in these nitrogen measurements corresponding to a POC variability of 88 and 31 mmol m⁻³, or an average of +60 for the river end-member shown in the bottom line of Fig. 3. This variability is relatively small, about +7 % of the POC signal, so that the calculations of POC removal given in the “Results” are fairly well constrained. The errors given for percent removed in the “Results” reflect the variability only in the measured POC concentration data, not the propagated error including end-members.

The offshore end-members in Fig. 3 were obtained by averaging results for high salinity (>35) samples, with 2008 and 2009 respective means + SEM (n) of 35.58 + 0.06 (5) and 35.49 + 0.16 (4) for salinity, 2,169 + 24 (5) and 2,134 + 41 (4) for mmol m⁻³ DIC, and 10 + 2 (5) and 16 + 5 (4) for mmol m⁻³ POC. The offshore end-members used in Fig. 3 were averages of these 2 years, 2,151 mmol m⁻³ DIC and 13 mmol m⁻³ POC at a salinity of 35.5.

For DIC data of 2008 and 2009 (Fig. 3), we used the AZE method critically reviewed by Regnier et al. (1998) to back-extrapolate the apparent 0 salinity river DIC end-member using the higher salinity data (open triangles in Fig. 3). The linear regression extending back from the high salinity data to 0 salinity had a DIC intercept value of 1,697 + 32 (standard deviation) mmol m⁻³, with the intercept error calculated in the statistical package in R (The R Project for Statistical Computing). This value is 59 mmol m⁻³ higher than the 1,638-mmol-m⁻³ value estimated above for complete nitrate + nitrite use and shown in Fig. 3 as the y-intercept of the middle line. However, this difference is small (only about 5 %) in terms of the total DIC drawdown near 1,100 mmol m⁻³ for these two 0 salinity scenarios where measured river DIC averaged 2,741 mmol m⁻³ for 2008 + 2009 (Fig. 3). The difference between 1,697 and 1,638 mmol m⁻³ may be regarded as not significantly different due to various errors, but also may reflect some net respiration of POC and DOC carried in by river water. According to decomposition experiments for potential lability, on average 17 and 38 mmol m⁻³ riverine POC and DOC, respectively, were metabolized in month-long experiments (Table 1). Respiration of these materials would return DIC to the shelf waters, elevating the 1,638 value 0 salinity value expected from complete nitrate + nitrite use to 1,693 mmol m⁻³, a value that is virtually identical to the measured 1,697-mmol-m⁻³ AZE value.