botomus papatasi, particularly in spring and summer when trans- mission intensity is highs; and viru- lent clones outcompete avirulent ones when this vector is experimen- tally infected with bothg. The relative abundance of virulent and avirulent isolates from gerbils changes dramatically with season. Figure la shows data collected from three regions of the Karshinskaya steppe between 1980 and 1984'O. Gerbils start to breed in early spring, and the newborn young remain unin- fected until transmission bysandflies begins in April (reaching maximum intensity in the following months). During this short interval, the preva- lence of infection in the gerbil popu- lation falls from its overwintering high of 70-100%. Transmission by P. papatasi is coincident with a rapid increase in the fraction of isolatesthat are virulent, possibly because viru- lent parasites have the advantage in this vector. Virulence and prevalence reach their peaks in late summer. Then, when transmission ceases, parasites surviving in overwintering animals become preponderantly avirulent again. transmitted (rather than vector- borne) infections; used in this con- text, it implies that the sandfly popu- lation varies with the square of gerbil population size. This may be approxi- mately correct because sandfly and gerbil breeding seasons coincide, and because sandfly numbers multi- ply many more times in a season than do gerbil numbers. Papers produced in the West'-5 have done much to clarify ideas about the trade-offs facing parasites and their hosts. The general problem for the parasite is that increased transmissibility will often have a cost in higher pathogenicity to the host. Higher pathogenicity means a higher host death rate and a shorter duration of infectiousness. For the host too, resistance to a new parasite genotype is unlikely to come cost free. Models that focus on the genetics or on the epidemiology, but rarely on both together5, have shown how the vari- ous trade-offs can lead to persistent polymorphisms.