ANTHER-SMUT INFECTION IN SILENE DIOICA - VARIATION IN FLORAL MORPHOLOGY AND PATTERNS OF SPORE DEPOSITION

The anther-smut Microbotryum violaceum, causes a systemic perennial infection of its host Silene dioica. Infection results in sterility and the production of teliospores in flowers. These spores are transmitted to healthy plants by flower visiting insects. The behavioural responses of flower visiting insects to a variation in floral characters are therefore likely to affect rates of pollen export/import, the rate of spore deposition and probability of disease. Since infected plants are sterilised, they are effectively removed from the gene pool. It is therefore often assumed that in this host-pathogen system there is a considerable potential for a pathogen-pollinator mediated selection on floral characters. In a transplantation experiment, which included plants from four different populations, we tested for correlations between variation in female floral morphology and patterns of spore and pollen deposition, and a resulting risk of disease. The source populations in this experiment were located on four islands in Skeppsvik archipelago in northern Sweden, and represented a gradient of disease incidence from completely healthy (Island 1), low incidence (Island 2) to high incidences (Islands 3 and 4) of disease. Fifty plants from each population were transplanted to the centre of the population on Island 4. Plants from the non-diseased population had larger flowers and longer styles than plants from the highly diseased populations. Numbers of pollen grains and spores deposited on flowers were strongly and positively correlated. We found that plants originating from the healthy population received approximately four times more pollen and nine times more spores per flower than the individuals from the resident diseased population. The resulting incidences of disease among plants from the four populations differed significantly, and was 20% among plants originating from the healthy population and 0% among plants from the resident highly diseased population. Plants from the other two diseased populations showed intermediate values (13% and 11% respectively). In a survey of ten populations we found among healthy plants a significant negative correlation between mean style length (positively correlated with corolla size and ovule number) and incidence of disease. The potential for pathogen-pollinator mediated selection on floral characters in this host-pathogen system is discussed.