Recent metagenomics studies indicate that crop-infecting viruses represent only a small fraction of a much richer plant virus diversity that includes many wild (non-crop) viruses found broadly throughout natural plant communities. However, essentially nothing is known about the dynamics of these wild plant viruses or their ecological impacts. To investigate the effects of landscape context, community diversity, and plant traits on rates of virus accumulation in the native prairie species Panicum virgatum (switchgrass), we established a network of experimental plantings within different Michigan landscapes and evaluated infection prevalence over time. At each site we grew a suite of different switchgrass communities in 400-m2 plots. Plantings included monospecific stands of five different P. virgatum cultivars, representing both lowland and upland ecotypes, and a diversity gradient in which P. virgatum was grown in monoculture, with other native grasses, and with other native grasses and forbs. After a three-year establishment period, P. virgatum in all communities was assessed for virus infection. Semi-purified virus concentrations from bulked survey samples were tested for the presence of multiple viruses and then the prevalence of specific viruses was assessed through RNA extractions and RT-PCR of individual samples with virus-specific primers.
Results/Conclusions
P. virgatum developed some infection with regionally common wheat viruses, as predicted. Unexpectedly, however, the dominant infection it acquired was that of a wild (non-crop) virus only recently discovered in prairie grasses. This virus -- Switchgrass mosaic virus (SwMV, genus marafivirus, family Tymoviridae) – is transmitted by leafhoppers and infects other native grasses including Andropogon gerardii and Sorghastrum nutans. Despite the disparate locations of our experimental sites, SwMV was present at all 12 sites, with mean site-level prevalence in 2015 ranging from <1 to 31%. Prevalence generally increased with time, reaching strikingly high values (>80%) in some plots. Within-plot community diversity influenced virus prevalence somewhat (P=0.001) but less than expected and much less than plant traits: Upland ecotypes with geographic origins closest to our sites accumulated considerably more infection than lowland ecotypes with more distant origins (P<0.001). Although it has been suggested that wild virus infections are benign or even beneficial, SwMV infection was associated with foliar chlorophyll loss in many cases. The discovery of widespread wild virus infection in native switchgrass raises urgent questions about its effects on host fitness and grassland dynamics.