These preliminary results are truly encouraging. But conclusions about the ori- gins, absolute genetic structure and identity of the DNA sequences must be considered as hypotheses that remain to be tested. The kind of data reported by Tyson and col- leagues represents a sort of average genome structure, a patchwork quilt combining pieces of DNA from many individual cells in a complex population. For instance, an estimated 100 million Leptospirillum-like cells contributed to a total of 29,000 assem- bled DNA sequence reads for this group. Each DNA sequence that was read probably originated from a different (and potentially non-identical) cell within the population. So the assembled scaffolds are very different beasts from the singular genome sequences (determined from clonal laboratory isolates) that currently populate genome-sequence databases. Such environmentally derived genome-sequence assemblies need to be very carefully flagged up in the databases, so that their distinction as composite population assemblies will be obvious. In the case of the Leptospirillum species, the nucleotide polymorphism rate (average sequence differences between individual reads) was low, about 0.08%. For this Lepto- spirillum population, then, there appeared to be only moderate genetic heterogeneity. In contrast, the Ferroplasma-like scaffolds showed a much higher polymorphism rate (around 2.2%) between individual reads that make up the scaffold. Given the locations and frequencies of these poly- morphisms, the authors guess that the Ferroplasma polymorphisms arise predomi- nantly from homologous recombination — a process whereby microbial cells can assimi- late large blocks of foreign DNA into their genome. There are,however,other plausible expla- nations for the patterns observed in the data. For example, the population may have origi- nally been much more complex, but recently have undergone a ‘selective sweep’, leaving behind highly related but non-identical polymorphic genotypes 6 . What Tyson et al. presume to be genomic types that have been variously blended by lateral gene transfer may, in contrast, simply represent the visible survivors (by linear descent) of complex lineages — the surviving nodes in a family tree that has been extensively pruned by natural selection. At present, the existing data cannot fully distinguish between these alternatives. Better quantitative data on the levels of heterogeneity within and between Ferroplasma populations in this habitat, and a more quantitative assessment of genotypic representation in the environment,may help to clarify the issue. But the complexities of deciphering genomic data from complex microbial populations are more of an opportunity than a difficulty 7 . After all, the reality is that genomes are dynamic biological structures: any notion that they are singular, unchang- ing entities does not capture the process that shapes the diversity we see today. It is only by peering directly into naturally occurring genomic diversity,as Tyson et al. 2 have done, that the tempo, mode and mechanism of genome evolution and diversification, and its relationship to higher-order biological and ecological processes, will become clear. Our currently static snapshot views of microbial genomic diversity have the poten- tial to develop into motion pictures as this emerging field develops. As predicted 8 , new vistas on the natural microbial world are opening up dramatically, in part owing to the new capabilities afforded by modern genomic technologies. ■ Edward F. DeLong is at the Monterey Bay Aquarium Research Institute, 7700 Sandholdt Road, Moss Landing, California 95039-0628, USA. e-mail: delong@mbari.org 1. Pace, N. R. Science 276, 734–740 (1997). 2. Tyson, G. W. et al. Nature 428, 37–43 (2004). 3. Edwards, K. J., Bond, P. L., Gihring, T. M. & Banfield, J. F. Science 287, 1796–1799 (2000). 4. Woese, C. R. Microbiol. Rev. 51, 221–271 (1987). 5. Fleischmann, R. D. et al. Science 269, 496–512 (1995). 6. Palys, T., Nakamura, L. K. & Cohan, F. M. Int. J. Syst. Bacteriol. 47, 1145–1156 (1997). 7. DeLong, E. F. Curr. Opin. Microbiol. 5, 520–524 (2002). 8. Woese, C. R. Microbiol. Rev. 58, 1–9 (1994). news and views 26 NATURE | VOL 428 | 4 MARCH 2004 | www.nature.com/nature D iscovered a hundred years ago, superconductivity describes the flow of electric current without resistance in metals. Most metals do not become superconducting until cooled to within about ten degrees of absolute zero. But the discovery, in 1987, of materials that become superconducting at much higher tem- peratures rekindled an old dream that one day room-temperature superconductivity might be achieved. One of the mysteries of the high- Superconductivity Turn up the temperature Piers Coleman A more elaborate picture is developing of what makes some materials superconduct at relatively high temperatures. With it come hints for how to design materials with still higher transition temperatures. Figure 1 Layering effect. a, A single layer of lanthanum copper oxide becomes superconducting at a temperature of 40 K, as the electrons in the layer order into pairs. b, If the number of copper-oxide layers is increased to three (with layers of insulating material between them), the temperature at which the material becomes superconducting rises to 130 K, because electron pairs are now able to ‘tunnel’ between the layers. c, But if the number of layers is increased beyond three, the transition temperature does not continue to rise — in fact, it falls. Chakravarty et al. 3 propose that the transfer of charge from the inner to the outer layers nucleates a second order parameter, in addition to the ordered pairing of electrons, and that this drives down the transition temperature. ©2004 Nature Publishing Group