standing the genetic and ecological processes that determine the structure and function of microbial metagenomes cannot but lead to new ways of describing patterns in Nature and could lead to the emergence of new theories integrating microevoutionary and macroevolutionary principles.

Viruses are important not only as pathogens, but also as agents of lateral gene transfer and catalysts that generate tremendous genetic variation in their specific hosts. Viral activity also has important consequences for turnover of the elements, for example, in carbon cycling in aquatic systems. It has only recently been recognized that virus particle numbers are enormous, often exceeding those of co-occurring cellular life. For example, seawater contains 10 times more bacteriophage than cellular microbes. Estimates suggest the biosphere harbors perhaps as many as 10³¹ viral particles (Edwards and Rohwer 2005). Given these vast numbers, the influence of viruses on biodiversity and evolutionary catalysis, and their role in biogeochemical cycling, there is considerable interest in characterizing naturally occurring virus populations. Metagenomics has recently provided an important avenue for exploring these ubiquitous and biologically important entities.

Of special interest is the recent evidence that viruses infecting marine cyanobacteria carry genes involved in photosynthesis (Lindell et al. 2004). Presumably that prolongs the lives of infected hosts (and thus increases virus yields), but another effect is to serve as a genetic bridge between different host species, coupling their evolution, at least as far as such genes are concerned.

Viruses present several unique and interesting opportunities and challenges for metagenomic analyses. Their numbers are large, their genomes are small, and their diversity is impressive. Viruses typically evolve rapidly, so gene sequence conservation is typically much less than that in cellular organisms. Practically speaking, although their numbers are great, their biomass is small, and cloning of viral genes has sometimes been problematic because of modified nucleotides and the cellular toxicity of some of their genes. Metagenomic methods, especially newer sequencing technologies that do not require cloning, may mitigate some of these problems.