2. At each site, fifteen grassland communities comprising four monocultures and eleven 4-species mixtures based on a wide range of species’ proportions were sown at two densities and managed by cutting. Forage species were selected according to two crossed functional traits, ‘method of nitrogen acquisition’ and ‘pattern of temporal development’.
3. Averaged across sites, years, and sown densities, annual weed biomass in mixtures and monocultures was 0.5 and 2.0 t DM ha-1 (7% and 33% of total biomass respectively). In over 95% of mixtures (across sites and years) weed biomass was lower than the average of monocultures, and, in about two thirds of cases, lower than in the most suppressive monoculture (transgressive suppression). Suppression was significantly transgressive for 58% of site-years. Transgressive suppression by mixtures was maintained across years and was independent of site productivity.
4. Based on models, predicted average weed biomass in mixture over the whole experiment was 52% less (95% confidence interval 30% to 75%) than in the most suppressive monoculture, and was significantly lower than the most suppressive monoculture across all mixtures and along functional trait axes. Transgressive suppression of weed biomass was significant at each year across all mixtures and for each mixture.
5. Weed biomass was consistently low across all mixtures and years and although, in some years, it increased with increasing sown legume proportion and decreased with increasing sown proportion of persistent species, the level of weed biomass was not largely different from that in the equiproportional mixture.
6. The average variability (standard deviation in t DM ha-1) of annual weed biomass within a site was much lower for mixtures (0.42) than for monocultures (1.77).
7. Synthesis and applications. Weed invasion can be diminished through combining forage species selected for complementarity and persistence traits in systems designed to reduce reliance on fertilizer nitrogen. The effects of mixtures on weed suppression were consistently strong across mixtures varying widely in species proportions and over time. The level of weed biomass did not vary greatly across mixtures varying widely in proportions of sown species. The benefits of plant diversity in intensively managed grasslands are highly relevant for the sustainable intensification of agriculture; importantly, this can be achieved through practical farm-scale actions.