Exclusionary interactions among diverse fungi infecting developing seeds of Centaurea stoebe.

Developing seeds are expected to be strongly defended against microbial attack. In keeping with this, only 26% of seeds of Centaurea stoebe from its native and invaded ranges in Eurasia and North America were infected with fungi, and 92.2% of those were infected with a single fungus per seed. Even when developing seeds in flower heads were inoculated under conducive conditions for infection with 14 of these seed-infecting fungi, re-isolation of inoculants was only 16% overall, and again limited to the particular inoculant. Environmental fungi (i.e. those not isolated from seed of C. stoebe) were present in control flower heads under conditions conducive to infection but they were never re-isolated from fully developed seeds in any experiments. When two or three seed isolates were co-inoculated to compete in flower heads, only one inoculant, and always the same one, was re-isolated from all matured seeds, regardless of maternal plant genotype. PCR-based detection methods confirmed that these fungal interactions were exclusionary rather than suppressive. In these strongly defended, developing seeds, we had expected the plant to control not only the overall level of infection but also the outcome of co-inoculations. Consequences for the next plant generation of this exclusionary competition among seed-infecting fungi included effects on seedling emergence, growth and fecundity.

Population-level compensation by an invasive thistle thwarts biological control from seed predators.

Pre-dispersal seed predators are often chosen as biocontrol agents because of their high impacts on plant fitness; however, they have a mixed record in realizing decreased plant population growth. Few studies have experimentally removed agents to explore their impact on weed population growth. Here, we used manipulative experiments with invasive yellow starthistle (YST), Centaurea solstitialis, and its pre-dispersal seed predator biological control agents, primarily Eustenopus villosus, the hairy weevil, and Chaetorellia succinea, the false peacock fly, to explore how these agents affect population growth of YST. We also use additional seed augmentation experiments to mimic effects of agents on seed inputs across a range of seed and adult plant densities. We found that biocontrol agents reduced seed production by > 70% and that seedling numbers were significantly related to seed inputs. However, several compensatory processes prevented effective population reduction of YST by seed predators. First, self-thinning reduced seedling numbers such that densities of plants in our agents-present and agents-absent treatments converged. Second, plots in which plants started at low density had particularly high population growth rates. In this case, plant plasticity and conservation of final yield, in which a small number of large plants produce as much seed as a large number of smaller plants occupying the same area, also provided avenues where plant populations can compensate for damage. Similarly, seed production on a per-plot basis was unchanged across a large range of YST densities. Our results suggest that at very low plant densities, biocontrol agents may reduce plant populations; however, other sources of mortality to YST (preferably imposed after self-thinning) will be needed to reduce populations to sizes where agents can become effective tools in weed control.