Efficiency of mechanical seed harvesting for grassland restoration

Maintaining and re-creating species-rich semi-natural grasslands are important issues in current agricultural policy in Europe. The seed that is required for their establishment can be obtained through direct harvesting from semi-natural herbaceous vegetation. To test the efficiency of mechanical seed harvesting on donor Arrhenatherion elatioris grasslands, experiments were performed in northern Italy and eastern Czech Republic. Trials were organized with a randomized block design and involved harvesting as green hay (grass mowing and immediate collection), dry hay (grass mowing and collection after drying on the field), direct combining (grass cutting and threshing at the same combine passage) and seed stripping with pull-type equipment (seed removal without grass cutting with a brush harvester pulled by a tractor). Harvesting was carried out at the time of maximum ripe standing seed yield (SSY) in the first and second regrowth. The harvested materials were analysed for seed number and weight and compared with the SSY. The species composition and phenology were also surveyed. In all methods, the seed mixture obtained contained the species present as seed at harvest time and was correlated with SSY. However, with regard to the seed number collected, the harvesting efficiency changed in relation to species group (grasses or forbs), individual species, seed maturation and regrowth. The most efficient method was harvesting as green hay (efficiency of approximately 71% of SSY and seed mixture composition that was very highly correlated with SSY). The least efficient methods were direct combining (30% of SSY and negative selectivity against light or difficult to detach seeds, producing the highest mean 1000-seeds weight) and harvesting as dry hay (39% of SSY and a high sensitivity to seed maturation level). Seed stripping, operated with downward brush rotation at the leading edge, resulted in an intermediate efficiency (60% of SSY and a seed mixture composition highly correlated with SSY). The harvesting efficiency for seed number was higher at the first regrowth than the second one for both species groups and, especially at the second regrowth, higher for forbs than for grasses. Especially forb harvesting presented a problem with regard to the species number collected. The problem was, not due to harvesting inefficiency but rather for phenological reasons, as several forbs did not produce fertile stems at the first or second regrowth and some other early flowering species had already shed the seed at the harvesting time. These results improve our understanding of factors affecting the efficiency of mechanical harvesting and will help in the preparation of efficient harvesting programs.