The implications of multilevel selection for virulence evolution deserve closer attention, especially where selection leads to conflicts of interest between organisms at different organizational levels. In pathogen-host interactions this conflict is selfevident, but in numerous cases parasites have evolved to act as commensalists or even as mutualists. The latter case, however, does not imply that interests exactly match.

In Chapter 20, Hoekstra and Debets consider mutants of mitochondria that slow down the growth of their host, a bread mold fungus. In spite of this apparent disadvantage, these mutants outcompete normal mitochondria in crosses between fungi that contain wild-type and mutant mitochondria. If such crosses occur frequently enough in nature, the resultant intragenomic conflict may lead to the interesting phenomenon that, through the lower-level selection process, fungal hosts with relatively slow growth can increase in frequency in the population. Similar processes can occur via mitochondrial plasmids that cause senescence, a phenomenon normally absent in fungi. The persistence of these obviously harmful plasmids is striking in some genera of fungi; the key to understanding this observation is probably the existence of horizontal transmission by anastomosis between different fungal units. Hoekstra and Debets suggest that, once horizontal transfer is open to manipulation, the performance of fungal diseases could be changed through intragenomic conflict.

Chapter 21 describes the evolutionary dynamics of the tritrophic interaction between chestnut trees, a blight fungus, and a double-stranded RNA virus.