Abstract1. Restoration of novel ecosystems to a historical benchmark may not always be possible or advisable. Novel ecosystems may be managed by targeting specific components and accepting the novelty of other ecosystem attributes. The feasibility of this component-wise management of novel ecosystems has rarely been tested.2. In a novel grassland, where C 3 grasses have replaced C 4 grasses, nutrients have been elevated, and diversity has been lost due to a history of agricultural land use, we aimed to return diversity using seed addition, without altering the dominant grass matrix or nutrient status. Using direct seeding, with and without soil disturbance, we assessed the ability of 10 species of native forbs to establish.3. Eight of the 10 seeded species established in the first year. Soil disturbance increased establishment success by 50%, while high levels of exotic cover decreased it by 24%. Establishment was inversely related to total plant cover at sowing, with a 10% increase in initial plant cover decreasing establishment by 47%.4. By the third year, six of the eight species persisted and five were flowering. Survival and reproduction in the third year was not associated with the soil disturbance treatment or plant cover. Synthesis and applications.We show that native plant species can be re-established in grasslands where abiotic and biotic conditions are novel relative to their reference state. This suggests that the conservation value of novel ecosystems can be enhanced using simple restoration tools that target specific ecosystem components.