Beneficial inherited symbionts are expected to reach high frequencies in their host population over time. However, most heritable symbionts are observed at intermediary frequencies in natural populations, including the presumed mutualistic fungal endophytes (Epichlöe spp.) in the grass host, Lolium multiflorum. Theory predicts that symbiont frequencies should reflect a balance of their benefits to host organisms and their efficiency of vertical transmission. To test the balance of these benefits, we established experimental plots with a range of initial endophyte frequencies in host populations and manipulated fitness benefits to hosts (by reducing drought stress) and vertical transmission. We analyzed the effects of water addition and endophyte status on plant demography, the effects of water addition on vertical transmission rate, and of effects of water addition and transmission reduction on the change in population level endophyte frequencies. Demographic data were measured over two generations and seeds were scored for fungal endophyte status.
We found that inflorescence production was significantly higher in both endophyte positive individuals and those in water addition plots, although there was no relative fitness benefit of endophyte symbiosis due to water addition. Individual level vertical transmission was not significantly different with water treatments and was high overall (~94%). Experimental reduction of vertical transmission significantly suppressed population-level endophyte prevalence, but water addition had virtually no effect. Our analysis revealed a stable intermediate prevalence of endophyte symbiosis, reflecting a balance of fitness benefits and imperfect transmission. Since water addition did not alter the fitness effects of endophytes, the advantage of symbiosis in this system appears unrelated to drought stress. Our novel approach of manipulating the environment and transmission processes provides the first experimental evidence supporting the role of imperfect vertical transmission in population level prevalence.