Five pasture treatments: (1) controlled grazing, (2) controlled grazing + lime, (3) controlled grazing + lime + phosphate, (4) controlled grazing + lime + phosphate + oversown white clover and (5) controlled grazing + lime + phosphate + oversown white clover + oversown perennial ryegrass were applied to three semi-natural hill grassland communities. The communities were those dominated by Agrostis/Festuca, Molinia and Nardus. The proportion of Nardus at the Nardus-dominant site was substantially reduced by herbicide before treatments were applied. All treatments were grazed simultaneously by mature wether sheep on three occasions each year. There were two grazing periods each of 4 weeks duration between mid-May and mid-August with a further grazing period of 3 weeks starting mid-October. During each grazing period stock numbers were set so that a residual herbage mass of 560 kg DM ha-1 remained at the end of the grazing period. Measurements of net herbage accumulation (NHA) were made annually over a period of 13 years at each site. The green:dead ratio of grasses, species composition of the pasture and the pH of the soil were measured at intervals during the experiment. Estimates of mean annual NHA ranged from 3860 kg DM ha-1 for treatment 1 (controlled grazing) to 5170 kg DM ha-1 for treatment 5 (oversown white clover and perennial ryegrass). (oversown white clover and perennial ryegrass). The application of lime and phosphate increased annual NHA by 300-350 kg DM ha-1 with a further increase of around 400 kg DM ha-1 when white clover was sown. Increases in NHA between year 1 and year 13 ranged from 30% for treatment 1 to around 55% for treatment 5. Although there was no difference in the mean NHA between sites, the herbage from the Agrostis/Festuca site contained a higher proportion of green grass and white clover than that from the other sites. The highest levels of green grass and white clover were found on the oversown treatments at each site. The grazing pressure exerted produced relatively little change in the botanical composition at the Agrostis/Festuca site. At the Molinia-dominant site the Molinia was largely replaced by Nardus during the first 6 years and Nardus also increased in cover at the Nardus site. Application of lime and phosphate generally increased the proportion of Poa pratensis, Festuca rubra and Agrostis tenuis but did not halt the spread of Nardus at the Molinia and Nardus sites. White clover and perennial ryegrass were successfully introduced by oversowing and proportions remained high throughout the 13 years. The cover by bryophyte spp. increased at all sites with the greatest increases occurring in each case on the less comprehensive pasture treatments. These results suggest that on Agrostis/Festuca-dominant pastures herbage biomass production can be increased with relatively low-cost pasture treatments while maintaining species diversity. However, Nardus and Molinia-dominant pastures are likely to require more comprehensive pasture treatments involving sown grasses and white clover to provide herbage of acceptable quality and avoid an increase in Nardus and bryophytes in the sward. With a regime of episodic summer grazing and the addition of fertilizers oversown pastures can be maintained over long periods of time.
