4.1SRB and MPA community structures revealed by membrane hybridization
The average SSU rRNA index of total SRB detected by all SRB probes used in this study was 2.3–5.1 and 20.0–22.9% of total SSU rRNA in the upper layers (0–2 and 2–8 cm) and the lower layers (8–14 and 14–20 cm), respectively (Table 2A). The values in the upper layers are comparable to those reported for other freshwater sediments. For instance, Gram-negative mesophilic SRB detected by probe SRB385 was 4.2% of all SSU rRNA detected by probe Univ1400 in the surface sediment (0–3 cm) in Lake Kizaki , and the combined SRB SSU rRNA was 1.6% of bacterial SSU rRNA in the freshwater site of the River Tama estuary .
In the upper sediment layer of Lake Biwa where sulfate reduction was relatively active (Fig. 2B), microorganisms detected by probe Dsv687 were most dominant in SRB SSU rRNA, followed by Desulfobulbus (Dbb660). Although the results were similar to those for freshwater lake sediments examined by the culture-dependent method [10,25], Desulfobulbus was also revealed to be predominant in freshwater sediments by the membrane hybridization method [23,24]. In any event, these results indicate that the incompletely oxidizing SRB contribute mainly to sulfate-reducing activity, resulting in the production of acetate in freshwater sediments. Maeda and Kawai  reported that acetate in the northern Lake Biwa sediment was predominant at all depths and had accumulated with the concomitant production of sulfides after the addition of various organic substrates.
The abundance of SSU rRNA hybridized to probe Dsv687 increased greatly at deep layers where no sulfate reduction was observed (Table 2A, Fig. 2B). A similar result has been reported by Sham et al.  for the Antarctic marine sediment in which the target microorganisms of this probe increased with depth to 36.0% of the bacterial SSU rRNA in the 25–28-cm layer. However, they could not detect Desulfovibrionaceae in the sediment either by MPN culture or a clone library . This probe hybridizes to some species of Geobacteraceae in addition to Desulfovibrionaceae[27,29]. The clones hybridized to the probe in their clone library were mostly related to Desulfuromonas palmitatis. All those clones, however, contained one mismatch, yet yielded a positive signal by the probe . Moreover, we could not obtain the sequences from Desulfovibrionaceae in the deep sediment of Lake Biwa by polymerase chain reaction-denaturing gradient gel electrophoresis (PCR-DGGE) analysis of reverse transcribed products from rRNA (Koizumi, unpublished data). Thus, the most abundant microorganism hybridized to this probe would not actually be SRB in the deep sediment. Further phylogenetic analysis such as by clone library will be needed to examine what kind of microorganisms were detected with probe Dsv687. However, other SRB SSU rRNAs (Desulfobulbus, Desulfobacterium and Desulfococcus) were still detected in the deep layers (Table 2A). These probes are highly specific as long as their specificities were evaluated by Probe_Match  and BLAST . Desulfobulbus and Desulfobacterium can survive by using alternative electron acceptors such as nitrate and Fe(III) or by fermentation under low-sulfate conditions . These results suggest that some kinds of SRB survive without sulfate in the deep sediments of freshwater lakes.
The sum of methanogens detected in this study (Msar860 and Mmic1200) accounted for approximately 40–70% of Archaea quantified by probe Arch915 (Table 2B). Archaea consists of three kingdoms, called Euryarchaeota, Crenarchaeota, and Korarchaeota. The distributions of Euryarchaeota other than mesophilic MPA and Korarchaeota are limited to only extremely salty or high-temperature environments [32,33]. Therefore, the remaining 30–60% of Archaea might be Crenarchaeota, since the wide diversity of this group in non-thermal freshwater environments has been reported in several studies [34–36].
SSU rRNA of the family Methanosaetaceae and the order Methanomicrobiales were the most abundant MPA among total archaeal SSU rRNA, accounting for 8.1–14.5 and 10.1–19.8% (20.5–27.6 and 18.8–30.6% of total MPA SSU rRNA), respectively. Falz et al.  detected two major clusters related to M. concilii and to a protozoan endosymbiont (the order Methanomicrobiales) in an archaeal clone library from the sediment of Rotsee (Switzerland). They also showed that Methanosaeta spp. represented on average 91% of the archaeal population by fluorescent in situ hybridization (FISH) using newly designed probes targeting the above clusters. Go et al.  also obtained 11 clones affiliated with Methanosaeta spp. along with 26 clones clustered around the order Methanomicrobiales from 40 archaeal SSU rDNA clones from the sediment of Lake Soyang (South Korea). The dominance of Methanosaeta was also consistent with a report on rice paddy soil . Methanosarcina was hardly detected throughout the layers in the present study, in accordance with the above reports. Most likely Methanosarcina is outcompeted by Methanosaeta in low acetate concentration environments since the acetate Km value estimated for Methanosarcina (barkeri) (about 5 mM)  is much higher than that for Methanosaeta (0.5 mM) . On the other hand, in Lake Michigan the dominant MPA were the order Methanomicrobiales and the family Methanococcaceae, whereas the order Methanosarcinales was less than 0.25% of total SSU rRNA throughout the sediment .
The sum of the two aceticlastic populations targeted with probe Msar821 and Msae825 did not reach the level of Methanosarcinales detected by probe Msar860 (Table 2B). There are non-aceticlastic genera in Methanosarcinaceae (e.g. Methanococcoides, Methanolobus, and Methanohalophilus), which utilize methanol, methylamines, and other C-1 compounds. However, these genera have been isolated so far exclusively from saline environments . Of six clones related to M. concilii obtained from the lake sediment by Falz et al. , four contained one mismatch to probe Msae825, but were perfectly matched to probe Msar860. A similar analysis of Methanosaeta-related clones obtained from other sediments also supports this observation [38,39,43]. Thus, Methanosaeta spp. with SSU rRNA containing few mismatches to probe Msae825 might dominate in freshwater sediments.
Although 11 probes specific for five SRB and six MPA were used for comprehensive understanding of the depth-related community structure changes of SRB and MPA, many more specific and selective probes are being developed based on the extensive SSU rRNA data bank. There is still possibility of existence of other kinds of SRB and MPA that could not be recognized by the probes used in this study. In order to choose probes for further species-specific analysis and to design new probes for uncultured groups, phylogenetic analysis (e.g. clone library) is needed.
4.2Specific cellular sulfate reduction and methane production rates
Vertical distribution patterns of SSU rRNA of SRB and MPA were different from those of SRR and MPR, although in general cellular rRNA content positively correlates with growth rate . The relationship between rRNA content and the cellular metabolic rate (e.g. SRR and MPR) may differ according to the bacterial species and growth conditions. Moreover, SSU rRNA recovery from sediment may not be constant. However, cellular SRR and MPR were roughly estimated in order to compare other studies and evaluate growth conditions. This estimation was conducted on the assumption that all microorganisms have the same abundance of rRNA. Cell numbers of SRB and MPA were estimated from the relative contribution of SRB or MPA rRNA to the total prokaryotic rRNA (Bact338 and Arch915) and direct cell counts derived from the sample of July 6, 1999. The cellular SRR was between 0.84 and 0.01 fmol cell−1 day−1, and decreased with depth (Fig. 3). The low value in the deep layer might be due to an overestimation of the Desulfovibrio proportion by probe Dsv687 as described above. These rates were comparable to the previous reports (0.01–0.09 fmol cell−1 day−1 (0–8.5 cm depth) by Sahm et al. , and 0.03–0.14 fmol cell−1 day−1 (0–10 cm depth) by Ravenchlag et al.  determined with the similar method.
Figure 3. Depth profiles of the apparent specific cellular SRR and MPR. SRR and MPR derived from the single sample of 6.7.1999 were used for calculation. Cell numbers of SRB and MPA were roughly estimated from the data of direct cell counts and the SSU rRNA index on 6.7.1999. The SSU rRNA indices for SRB and MPA were the summation of ratios of specific SRB probe signals (Dsv687, Dsbb660, Dsbm221 and Dscoc814) and MPA probe signals (Msar860 and Mmic1200) to the total of prokaryote target probe signals (Bact338 and Arch915), respectively.
Download figure to PowerPoint
The cellular MPR was negligible at 0–2 cm but relatively constant at about 0.2 fmol cell−1 day−1 below 2 cm. These rates are much lower than the previous data  obtained from pure culture experiments, manure digestor, bovine rumen, and salt marsh sediment based on MPN enumeration (480–960, 24–72, 384, and 2.4–7.2 fmol cell−1 day−1, respectively). The apparent cellular MPR of the pure culture and in bovine rumen were approximately 10 000 times higher than the rate in the present freshwater sediment because these rates were obtained under nearly optimum conditions, with abundant substrates and optimum temperature (30–37°C). In addition, culture-dependent cell counts seem to underestimate MPA cell numbers. The accumulation of such data would be important in estimating the number and contribution of SRB and MPA in natural environments by culture-independent methods like SSU rRNA membrane hybridization.