ficial even on functional grounds. For example, when they are in bacteria, the prophage of the phage P1 is a plasmid and that of Mu is a transposon. Moreover, although some plasmids and a few transposons code for the machinery needed for their own infectious transfer by conjugation, the mobility of many plasmids and of most transposons, integrons, islands, and even ordinary genes is by hitchhiking on conjugative plasmids or phages or by being picked up as free DNA by hosts with transformation mechanisms. Although the modes of replication and transmission are critical for considerations of the population dynamics, existence conditions, and ecology of these different classes of accessory elements (Campbell, 1961; Chao and Levin, 1981; Condit et al., 1988; Levin and Stewart, 1980; Stewart and Levin, 1977, 1984), they are of only secondary import for a general consideration of adaptive evolution in bacteria. From this perspective, the most important factor is the extent to which these elements are mobile and the range of bacterial hosts they infect. In this sense, the accessory bacterial genetic elements can be seen as arrayed along a continuum from phages, plasmids, and transposons to pathogenicity (and nicer) islands, integrons, and even stay-at-home chromosomal genes.
Accessory elements at the most mobile end of the continuum may be maintained as “genetic parasites,” spreading by infectious transfer alone without bearing genes that augment the fitness of their host bacteria. A parasitic existence is almost certainly the case for purely lytic (virulent) phages (Levin et al., 1997) and possibly for many temperate phages as well (Stewart and Levin, 1984). Although a formal possibility, we believe that it is unlikely that plasmids and transposons are purely parasitic (Levin, 1993) and even less likely that islands and integrons are maintained without at least occasionally paying for their dinner. The rates of infectious transfer of these elements are almost certainly not great enough to overcome the fitness burden their carriage imposes on their host bacteria (Bouma and Lenski, 1988; Levin, 1980; Modi and Adams, 1991) and their losses by vegetative segregation. If this assumption is correct, then these elements must bear genes that are at least sometimes beneficial to their host bacteria. After all, these accessory elements are vertically transmitted (in the course of cell division); thus, it would be to their advantage to carry genes that augment the fitness of their hosts. But can this “niceness ” account for the maintenance of accessory genetic elements? If accessory element-borne genes provide a selective advantage to bacteria and the accessory elements themselves are either costly or unstable, why are those genes not sequestered by the host chromosome?