Notes: Silicified deposits, such as sinters, occur in several modern geothermal environments, but the mechanisms of silicification (and crucially the role of microorganisms in their construction) are still largely unresolved. Detailed examination of siliceous sinter, in particular sections of microstromatolites growing at the Krisuvik hot spring, Iceland, reveals that biomineralization contributes a major component to the overall structure, with approximately half the sinter thickness attributed to silicified microorganisms. Almost all microorganisms observed under the scanning electron microscope (SEM) are mineralized, with epicellular silica ranging in thickness from 〈 5 μm coatings on individual cells, to regions where entire colonies are cemented together in an amorphous silica matrix tens of micrometres thick. Within the overall profile, there appears to be two very distinct types of laminae that alternate repeatedly throughout the microstromatolite: ‘microbial’ layers are predominantly consisting of filamentous, intact, vertically aligned, biomineralized cyanobacteria, identified as Calothrix and Fischerella sp.; and weakly laminated silica layers which appear to be devoid of any microbial component. The microbial layers commonly have a sharply defined base, overlying the weakly laminated silica, and a gradational upper surface merging into the weakly laminated silica. These cyclic laminations are probably explained by variations in microbial activity. Active growth during spring/summer allows the microorganisms to keep pace with silicification, with the cell surfaces facilitating silicification, while during their natural slow growth phase in the dark autumn/winter months silicification exceeds the bacteria’s ability to compensate (i.e. grow upwards). At this stage, the microbial colony is probably not essential to microstromatolite formation, with silicification presumably occurring abiogenically. When conditions once again become favourable for growth, recolonization of the solid silica surface by free-living bacteria occurs: cell motility is not responsible for the laminations. We have also observed that microbial populations within the microstromatolite, some several mm in depth, appear viable, i.e. they still have their pigmentation, the trichomes are not collapsed, cell walls are unbroken, cytoplasm is still present and they proved culturable. This suggests that the bulk of silicification occurred rapidly, probably while the cells were still alive. Surprisingly, however, measurements of light transmittance through sections of the microstromatolite revealed that photosynthetically active light (PAL) only transmitted through the uppermost 2 mm. Therefore the ‘deeper’ microbial populations must have either: (i) altered their metabolic pathways; (ii) become metabolically inactive; or (iii) the deeper populations may be dominated by different microbial assemblages from that of the surface. From these collective observations, it now seems unequivocal that microstromatolite formation is intimately linked to microbial activity and that the sinter fabric results from a combination of biomineralization, cell growth and recolonization. Furthermore, the similarities in morphology and microbial component to some Precambrian stromatolites, preserved in primary chert, suggests that we may be witnessing contemporaneous biomineralization processes and growth patterns analogous to those of the early Earth.

Notes: The sequence 5′-GCGATCGC- 3′, designated HIP1, for highly iterated palindrome, was first identified at the borders of a gene-deletion event and subsequently shown to constitute up to 2.5% of the DNA in some cyanobacteria. It is now reported that HIP1 is polyphyletic, occurring in several distinct cyanobacterial lineages and not defining a clade. HIP1 does not introduce gaps into sequence alignments. It aligns with partial HIP1 sites in related sequences showing that it propagates by nucleotide substitutions rather than insertion. Constructs have been created to determine the frequencies at which deletion events occur between palindromes located within the selectable marker neoDeletion between HIP1 sites was more frequent in Synechococcus PCC 7942 than deletion between control palindromes, 5′-CCGATCGG-3′, designated PAL0. However, this is not due to a recombinase that recognises HIP1 and is peculiar to cyanobacteria because similar deletion frequencies were detected in Escherichia coli. Furthermore, the frequency of deletion of DNA flanked asymmetrically by one HIP1 site and one PAL0 site was less than the frequency of deletion of DNA flanked symmetrically by identical copies of either palindrome. This is consistent with deletion by copy-choice.