From the Cover: Quantifying archaeal community autotrophy in the mesopelagic ocean using natural radiocarbon -- Ingalls et al. 103 (17): 6442 Data Supplement - HTML Page - index.htslp -- Proceedings of the National Academy of Sciences

Ingalls et al. 10.1073/pnas.0510157103.

Supporting Information

Files in this Data Supplement:

Supporting Figure 3
Supporting Figure 4
Supporting Figure 5
Supporting Text

Fig. 3. Surface sample. HPLC/atmospheric pressure chemical ionization(APCI)-MS total ion current chromatograms of the GDGT fractions after two-dimensional (2D) separation, as submitted for analysis by ¹⁴C-AMS. (A) Total ion chromatogram of the bulk GDGT lipid sample. (B) Separated fraction containing GDGTs I, III, IV, and V with m/z 1302, 1300, 1298, and 1296, respectively. Inset is a mass spectrum of the total purified fraction. (C) Separated fraction containing GDGTs II and VI, both with m/z 1292. Inset is a mass spectrum of the total purified fraction.

Supporting Figure 4

Fig. 4. Deep sample. HPLC/APCI-MS total ion current chromatograms of the GDGT fractions after 2D separation, as submitted for analysis by ¹⁴C-AMS. (A) Total ion chromatogram of the total sample. (B–F) Chromatograms of each separated GDGT. Insets are mass spectra of the total purified fractions. (B) GDGT I, m/z 1302. (C) GDGT III, m/z 1300. (D) GDGT IV, m/z 1298. (E) GDGT II, m/z 1292. (F) GDGT VI, mass 1292.

Supporting Figure 5

Fig. 5. Model optimization for pairs of f_S and D_D, plotted against the absolute values of the differences between the modeled and measured values for D_Ti (weighted relative to the size of the measurement errors). The bold regions of the lines indicate solutions of the model in which all calculated values of D_Ti agree with the actual measured values of D_Ti within the 1s measurement errors of the actual data. The best solution is the minimum total combined error that falls within this region, and it is indicated with an arrow at
–112‰.

Supporting Text

Assessment of D¹⁴C Blanks and Error Corrections.
To determine the blank associated with the HPLC purification method, blanks containing 6, 12, 18, 36, 72, and 108 ml of HPLC effluent from the Agilent ZORBAX Eclipse XDB-C₈ column were collected and combusted exactly as samples of individual compounds. A combustion blank of 1.0 ± 0.2 mg•C and a HPLC-derived blank of 0.03 ± 0.01 mg•C/ml effluent was calculated from a regression line drawn from a plot of the mass for each blank vs. volume of effluent. The combustion blank is significantly larger and more variable than the HPLC contribution. Six stigmasterol standards (Sigma Lot no. 47H5033), ranging from 2.5 to 12.5 mg•C, also were collected using the same HPLC program. Each standard was combusted, and CO₂ was captured as described above. One milligram of pure stigmasterol standard also was directly transferred to a 9-mm quartz tube, dried, amended with »2 g of cupric oxide, evacuated, and combusted for 5 h at 850°C.

D¹⁴C measurements were made for the sample containing 108 ml of HPLC blank and for the stigmasterol standards at the Keck Facility, University of California, Irvine (UCIAMS). Treating the D¹⁴C value of the 1-mg stigmasterol sample as the true value, and using the quantities of combustion blank and HPLC blank calculated above, the D¹⁴C values of the combustion blank and the HPLC blank are 21 ± 215‰ and -755 ± 66‰, respectively. The uncertainty associated with the D¹⁴C of the combustion and HPLC blanks was calculated using partial-derivative methods (J. M. Hayes, www.nosams.whoi.edu/docs/IsoNotesAug02.pdf).

Other potential sources of contamination during preparation of compound-specific ¹⁴C samples include blanks associated with flame-sealing of quartz and Pyrex tubes, blanks associated with use of vacuum lines, contamination on interface devices (tube crackers and fittings), and blanks associated with the portions of the outsides of the tubes that must be exposed to the interior of the vacuum environment. Each of these possible sources of additional carbon is expected to contribute a constant mass per sample. After correcting the individual compound data for the (measured) combustion and HPLC blanks, the D¹⁴C data still contained a strongly mass-dependent (R² = 0.92 for deep archaeal lipids) component of residual carbon when the D¹⁴C data were plotted as a function of 1/m. This indicates the residual blank had a constant mass and D¹⁴C value and probably was derived from one or more of the sources described above. Using the assumption that these unmeasurable contaminants are most likely to be radiocarbon-dead (D¹⁴C = –1000‰), the total quantity of extra contaminant carbon was calculated to be 0.7 mg•C; this is smaller than the combustion blank and is a reasonable result. Correction for this component was made to the values reported in Tables 1 and 2.

The reported uncertainty associated with the AMS facility processes (graphitization, handling, AMS counting statistics, and analysis corrections for small mass samples) always was smaller than the total propagated uncertainty associated with the measurements. This calculation highlights the importance of propagation of error when dealing with extremely small samples. The total propagated error was calculated by re-deriving the partial derivative equations 5.10–5.19 originally defined by J. M. Hayes (www.nosams.whoi.edu/docs/IsoNotesAug02.pdf) but here adapted for the case of a sample having three blank corrections.