RESULTS

Examination of the filters with the electron microscope revealed that the filter fibers were peppered with particles that were less than 1 µm (Fig. 3A). Rod-shaped particles that were less than 0.5 µm were typical, as were spheres (Fig. 3A). These particles have the morphology of cells, and in several cases appear to be joined end-to-end. However, they are possibly too small to be bacteria. Alternatively, they could be minerals entrained in the subsurface water or that precipitated from the water after it was collected. Particles larger than 1 µm were also present on the filter. These could have been already present in the WSTP when it was deployed or they could have passed through the 1-µm filter on the WSTP. A clay particle on the filter is shown in Figure 3B. This clay particle does not have the coating of small particles that were seen in Figure 3A. The clay particle is resting on unidentified material (which we named "matte") that also was collected on the filter (Fig. 3B). Matte may be polysaccharide, which appears to be abundant in the waters of Hole 1026B (S. Giovannoni, unpubl. data). In addition to mineral grains, there are a few objects that appear to be of biological origin (Fig. 3C). These are similar to what has been described as microbial floc (Juniper et al., 1995) from the Juan de Fuca Ridge, which were 20%-25% Fe.

Particles on the filter were analyzed with an energy dispersive spectrometer attached to the SEM using a standardless technique. The analyses were normalized to 100% and should be considered qualitative (Table 1). In addition to those elements listed in Table 1, analysis of phosphorus was attempted, but it was not detected. The particles were irregular, which also affected X-ray absorption and fluorescence. Cu was detected in all of the analyses, and in some of them Cu is the major component. This is not an artifact of the analytical technique, because Cu was detected with the other microbeam analyses. Cu is present in the 0.45-µm Acrodisk filters, but at a level of only 2 ppm, so this is not the source of the Cu signal.

Cu is inversely correlated with silicon and aluminum; Si and Al have a roughly constant ratio (Fig. 4A). This suggests that the analyses are a mixture of clay and a copper compound. The makeup of the copper compound is not obvious from the analyses. Mg and Cu appear to be strongly correlated (Fig. 4B), but Cu is not strongly correlated with any other metals, sulfur, or chlorine (Fig. 4C). Fe, Cr, and Ni were elevated in some analyses, suggesting that rusted steel from the drill pipe could have contaminated the sample.

Additional EDS and wavelength dispersive analyses of particles on a filter from Run 2 of the WSTP appear to be clays, oxides, and sulfates (Table 2). Euhedral crystals of calcium sulfate appear to have formed on the filter when they were dried. Cu was observed in the EDS spectra of particles on the filter as was found with EDS analyses in Bergen.

A TEM image of the filter is shown in Figure 5. In addition to the particles shown in the figure, qualitative chemical identification (using the energy spectra from the energy dispersive detector) was made of particles elsewhere on the filter (Table 3). TEM spectral analyses always had Cuk, CuK_ß, CuL, Ck, and SK peaks. The filter and background were also measured. As in the SEM EDS analyses of the particles (Table 1), Cu is ubiquitous. The filter fiber showed the presence of S, Cu, and trace amounts of Si. Particles trapped on the filter included matte, aluminosilicates (presumably clay), steel, titanium-rich grains, sulfur-silica-iron-rich particles, and barite. A single analysis of what we have called a cell (Fig. 5) did not appear to be significantly different from the matte, except that the cell had less iron. The matte was interesting in that it was rich in C, O, Si, Fe, and Cu, as well as P. One Ti-rich particle was analyzed as well as one barite. Anatase (TiO₂) has been reported as an extreme weathering product of basalt (Howard and Fisk, 1988) and could be carried in the subsurface fluids; however, it could also be derived from the titanium pressure casing of the overflow chamber.

Fluorescence caused by the dye Syto-59 was localized on particles on the filter (Fig. 6). Reflection from the particles are seen in green and the fluorescence is indicated by red. Nucleic acids on the particles are indicated by red isolated pixels that are typically located in 1-µm spots throughout the particles.

Amplification with domain-specific 16S rRNA PCR primers failed to detect either bacterial or archaeal sequences in DNA fractions from a Leg 168 particulate sample. Amplification products were detected only in positive control samples containing DNA extracted from E. coli spotted onto filters and in positive controls containing E. coli DNA extracted by standard protocols (a faint band was also present in one of the negative control filter extract as well). The most likely explanation for the failure of this effort is that there were too few cells present on the filter to be detected by PCR. However, another possible explanation for our inability to detect 16S rRNA genes from the Leg 168 sample is that a PCR-inhibitory substance was extracted from mineral particles present on the sample filter. Although we could test for this possibility by adding purified, positive control template DNA to Leg 168 filter extracts and assaying by PCR, we have too few Leg 168 samples to troubleshoot a purification protocol and eliminate such an inhibitory substance.