Spontaneous recovery of an intensively used grassland after cessation of fertilizing

Questions (1) Does a fertilized grassland recover in terms of species composition and species richness after stopping the treatment? (2) Which species contributed most to the changes? (3) Is it realistic to rely upon spontaneous recovery in this and similar cases? Location A large grassland, located in the western part of the Czech Republic in central Europe, latitude 49 degrees 59', longitude 12 degrees 45', altitude 750 m a.s.l. Methods Experimental plots of 2.5 m x 6.0 m were established in the central part of the grassland in 1968 using a randomized block design with the following variants: unfertilized controls, 80, 160, 240 and 320 kg ha(-1) of nitrogen, in four or eight (320 kg) replicates. The fertilization regime was stopped in 1989. Phytosociological releves of 2 m x 4 m in size were made in 1990, 1994, 1996, 1997, 1999, 2003 and 2006. The data were processed by multivariate ordination methods (DCA, CCA). Differences in the number of species per plot were tested by repeated measures anova and Tukey's HSD test. Results All plots, including controls, exhibited a distinct trajectory through time, reflecting the rates of fertilizer used and generally changing in the same direction. Both time and nutrients significantly influenced the vegetation pattern. Treatments differed significantly in species richness, especially at the beginning of the observation. Differences in later years were less pronounced, thus a convergence between the treatments was evident. Conclusions During the 16 years after fertilization ceased, the number of species substantially increased not only in the fertilized plots but also in the controls. Species typical of semi-natural grasslands predominantly contributed to the increase. Spontaneous recovery after cessation of fertilizing was relatively efficient, but the fertilized plots were small in size and thus rather easily colonized from outside. For interpretation and extrapolation of results from small experimental plots, it is necessary to consider changes in the broader surroundings.