Epichloae infection in a native South African grass, Festuca costata Nees

Abstract

Fungal endophytes have been documented in almost all terrestrial plant groups. Although the endophyte infection syndrome in agronomic cultivars is well studied, relatively little work addresses questions of spatial ecology and fire effects on epichloae endophyte infection in native grasses, and none, to our knowledge, in sub-Saharan Africa. We sampled seven populations of the native Festuca costata Nees along the spline of the Drakensberg range in South Africa at several spatial scales, including both recently burned and unburned stands. We tested epichloae presence and prevalence with immunoblot assays, PCR and genetic sequencing. We found epichloae endophytes were present and prevalent (38-98% infection rates depending on location). Variation in infection rates occurred primarily among locations, but also among bunches. There was little evidence that endophyte infection rates varied with fire. Novel evidence of epichloae infection of a native Festuca in South Africa opens the door to several new research questions, from the phylogenetic relationship between epichloae of sub-Saharan Africa and other continents to the ecological advantages or disadvantages that endophytes confer upon their hosts, especially in a fire-prone ecosystem vulnerable to global environmental change.

Temporal variability in aboveground plant biomass decreases as spatial variability increases

“…We modified the custom function RE.var (McGranahan et al. ) to extract variance components from a random‐effect regression model fit to a Gaussian distribution with the lmer function in the R package lme4 (Bates et al. , R Development Core Team ).…”

Temporal variability in aboveground plant biomass decreases as spatial variability increases

“…Our model included a temporal term to represent variation in aboveground biomass across years (gamma variability) and a spatial term to represent variation in aboveground biomass among local communities (beta variability). We modified the custom function RE.var (McGranahan et al 2015) to extract variance components from a random-effect regression model fit to a Gaussian distribution with the lmer function in the R package lme4 (Bates et al 2013, R Development Core Team 2013. See the Supplement for script for RE.var and all statistical analyses.…”