The firefly Luciferase assay is affected by pH, ionic strength, metallic ions, and the presence of chromophores in the extracts. Hence, the content of ATP was determined from standard addition curves as recommended by Ciardi and Nannipieri (1990). All standard addition curves had linear regression coefficients higher than 0.995. Although treatment with the cation exchanger and the use of the HEPES buffer reduced some of these interferences, considerable intersample variations in the specific bioluminescence (luminescence/mass of ATP) remained. The slope of the standard addition curves varied by a factor of 2.5 between the extreme cases (Fig. 2A). Despite these variations, the mean relative standard deviation between single point determinations of ATP and determinations of ATP based on the standard addition curves was only 8.6% (maximum relative standard deviation = 28%; Fig. 2B).
The sedimentary concentrations of ATP (360-7050 pg g-1, mean 1460, n = 54; Table 1) were in the expected range (see "Introduction" section, this chapter). The sampled sequence may be divided in two parts based on the ATP results: an upper part (0-30 mbsf) where the concentrations of ATP fluctuated considerably, and a deeper part (30-58 mbsf) with a remarkably constant level of ATP (Fig. 3). The distribution of ATP concentrations in the upper part of the sediment did not show any obvious depth trend.
A rough idea of the number of bacterial cells may be obtained by the use of literature data on C/ATP and C/bacteria cell ratios. Data collected by Karl (1980) show that the C/ATP ratios of microbial cells vary considerably (28-510, Karl [1980], table 3, p. 752). Despite these significant variations, indirect data from several independent field investigations have indicated that the commonly used mean ratio of 250 is a close approximation of the true in situ C/ATP ratio (Karl, 1980). The minimum and maximum ATP concentrations corresponded to 10-7 and 2 × 10-8 cells g-1 of sediment respectively assuming a conversion factor of 10-14 g C cell-1 (Gerlach, 1978; Findlay et al., 1986). This factor is well within the range of published conversion factors, however it is somewhat conservative and may underestimate bacterial carbon (Schallenberg and Kalf, 1993). Nevertheless, conversion to cells per cubic centimeter resulted in a surprisingly good overall match with the direct bacterial count numbers from Wellsbury et al. (Chap. 36, this volume) (Fig. 4). Much of the variations in the ATP-based bacteria numbers may be attributed to geochemical factors (see "Correlation of ATP and Geochemical Parameters" section, this chapter); hence the smoother trend of the bacteria numbers based on direct enumeration may be due to the limited numbers of observations.
None of the samples contained particles greater than 65 µm. All samples were characterized by a bimodal particle-size distribution (Fig. 5), with one peak at 1.8 µm and one peak at 9.8 µm. The positive correlation (see "Correlation of ATP and Geochemical Parameters" section, this chapter) between CaCO3 and mean particle diameter suggests that carbonate skeletons contribute to the population of larger particles, whereas the population of smaller particles probably consist mainly of clay minerals. On number basis, the mean particle diameter varied between 0.96 and 0.58 µm, and there is an inverse relation between mean particle diameter and sediment porosity (Fig. 3).
The content of CaCO3 and organic C ranged between 7.2%-57.2% and 0.14%-1.33% by weight, respectively (Table 1), and compared favorably with the shipboard determinations (Fig. 3). The analytical program used in this work provides two measurements of total N, one on the bulk sample when analyzing for total C and one on the acid-washed sample when analyzing for organic C. After correcting the latter for the loss of weight caused by dissolution of CaCO3 (assuming 100% dissolution of calcium carbonate), the two N determinations yielded very similar results (mean relative standard deviation of 4.5%, n = 54; Table 1). The mean concentration of N ranged from 0.035 to 0.126 wt%, whereas the range of shipboard determinations of total N on samples from the same interval is 0.0005-0.064 wt%. The C/N values estimated from the shipboard data (16-790) indicate that the shipboard determinations of N are unreliable for this part of the sediment sequence. The range of C/N values (4.6- 16) based on results from this work indicate that the organic matter is of a mixed terrestrial and marine origin. Similar results were noted by Katz (1983) in sediment collected from the Blake Ridge during Deep Sea Drilling Project (DSDP) Leg 76.
Principal component analysis is aimed at finding and interpreting hidden, complex, and possibly casually determined, relationships between features in a data set. The method decomposes the variance of a data matrix into uncorrelated (orthogonal, independent) principal components (PCs), which are linear combinations of the original variables. Interpretation of a PC is made possible by examination of the sign and magnitude of the loading in loading plots.
The results (Fig. 6) reveal several interesting features that may help explain factors governing the distribution of ATP (and by inference, bacterial numbers) in the sediments collected from Site 994. The two first PCs explain 63% of the variance, and there is a distinct grouping of the geochemical parameters. The positive correlation of organic carbon, total nitrogen, and porosity on PC1 is expected because most of the nitrogen is organic (adsorbed inorganic nitrogen contribute little), and fine-grained, high-porosity sediments usually contain more organic debris than coarse, low-porosity sediments. The fact that the C/N ratios also correlate with organic carbon suggests that enrichment in organic carbon is caused primarily by terrestrial organic matter (Jasper and Gagosian, 1990; Matson and Brinson, 1990). Distinctive sedimentary features clearly indicative of contour-current deposition of long-distance transported continental slope sediments were observed in lithologic Units I and II (0-160 mbsf; Paull, Matsumoto, Wallace, et al., 1996). Old, reworked terrestrial organic debris is less susceptible to bacterial breakdown, and this is probably why ATP is almost uncorrelated with organic carbon and total nitrogen (Fig. 6). It has been noted previously that organic matter quality may be more important than organic matter quantity (e.g., Parkes et al., 1990), although on a broader scale it is firmly established that organically enriched environments in general contain higher bacterial numbers than organic-poor environments (e.g., Schallenberg and Kalf, 1993).
The largest loading of ATP is found on PC2, which also contains high contributions from CaCO3 and particle diameter. One possible interpretation is that the content of CaCO3 particles reflects input of marine organic matter that is younger and more available for microbes than terrestrial organic matter, but that the amount of organic carbon associated with the carbonate skeletons is insufficient to impact the C/N ratio. It may also be speculated that particle size is the primary governing factor. The mean average particle diameter (0.58-0.96 µm), and hence the size of the pore space, is in the same range as the size of sedimentary bacteria (Schallenberg and Kalf, 1993). Therefore, the number of bacteria may be limited by spatial restrictions. Sediments enriched in coarse carbonate skeletons would promote proliferation of bacteria due to larger pore spaces and the supply of marine organic matter. In shallow-water sediments, bacteria numbers have been found to correlate negatively with mean grain diameter because coarse sediments are depleted in organic matter (Al-Rasheid and Sleigh, 1995). However, the mean grain diameter of these sediments were 2-3 orders of magnitude greater than the sediments studied here.
Inclusion of pore-water data on sulfate and methane concentrations did not enhance the amount of variance explained by PC analysis. This should not be surprising because sulfate-reducing bacteria and anaerobic fermentative heterotrophs constitute only a small fraction of total bacteria numbers in deep-sea sediments (e.g., Cragg et al., 1992; Cragg et al., 1996).