Recently it has been suggested that ploidy level of a plant population may have important effects on plant-animal interactions. Plant-animal interactions can also be strongly altered by factors such as plant population size and habitat conditions. It is, however, not known how these factors interact to shape the overall pattern of plant-animal interactions.
I studied the interaction between a perennial plant, Aster amellus, and a monophagous herbivorous moth, Coleophora obscenella, and investigated the effect of ploidy level of the plant population, plant population size, isolation and habitat conditions on density of the insect, damage by the insect, and plant performance.
Ploidy level, plant population size and habitat conditions, but not isolation, strongly influence plant-herbivore interactions. Furthermore, there are significant interactions between effects of ploidy level and plant population size and between ploidy level and isolation. Hexaploid plants suffer higher seed damage by the herbivore, but their seed production is still higher than that of diploids. Herbivores thus partly limit the evolutionary success of the hexaploid plants. Plant-animal interactions are also strongly determined by plant population size. Small populations of A. amellus (below forty flowering ramets) host no C. obscenella larvae, indicating a minimum A. amellus population size that can sustain a viable C. obscenella population. Negative and positive effects of plant population size balance and result in no relationship between plant population size and number of developed seeds per flower head. The results also show a significant interaction between ploidy level and plant population size, indicating that the increase in density of C. obscenella larvae with plant population size is greater in hexaploid than in diploid populations. The results also indicate that the effect of ploidy level on plant-herbivore interactions can be altered by plant population size, which suggests that plant-herbivore interactions are driven by a complex of interactions among different factors. Studying each factor separately could thus lead to biased conclusions about patterns of interactions in such systems.