Priority effects provide an advantage to early establishing species and are thought to significantly affect the course of succession. We conducted a 20-year long experiment sowing high- and low-diversity mixtures in an ex-arable field. We ask how long the effect of sowing persists and which sown species affect the course of succession. The experiment was established in the Czech Republic in five replicate blocks, each containing three random 10 × 10 m plots with three treatments: natural colonisation, sowing low- and high-diversity seed mixtures. The species cover was annually estimated in 12 permanent 1 m quadrates within each plot. To identify the effects of sowing, we used an innovative method analysing the data separately for each year using Redundancy analysis (RDA) with identity of sown species as explanatory variables. In the first year, the effect of sowing was small; the peak of explained variability occurred between third and fifth year. The legacy of sowing was detectable in the natural colonisers for 18 years and in the sown species for the whole 20-year period. For some species, the difference between the plots where they were and were not sown remained significant for the whole 20-year period (e.g. Lathyrus pratensis) although the plots were adjacent and the area was mown with the same machine. Other ones (e.g. Trisetum flavescens) colonised all the plots evenly. The long-lasting effect of the initial sowing confirms contingency of successional pathway on the propagule pressure in the time of start of succession due to the priority effects.
