A hierarchical model for analysing the stability of vegetation patterns created by grazing in temperate pastures

Questions: Does vegetation structure display any stability over the grazing season and in two successive years, and is there any correlation between the stability of these spatial patterns and local sward composition? Location: An upland grassland in the French Massif Central. Method: The mosaic of short and tall vegetation stands considered as grazed and ungrazed patches respectively is modeled as the realization of a Boolean process. This method does not require any arbitrarily set sward-height thresholds to discriminate between grazed and ungrazed areas, or the use of additional variables such as defoliation indexes. The model was validated by comparing empirical and simulated sward-height distributions and semi-variograms. Results: The model discriminated between grazed and ungrazed patches at both a fine (1 m(2)) and a larger (500 m(2)) scale. Selective grazing on legumes and forbs and avoidance of reproductive grass could partly explain the stability of fine-scale grazing patterns in lightly grazed plots. In these plots, the model revealed an inter-annual stability of large-scale grazing patterns at the time peak biomass occurred. At the end of the grazing season, lightly grazed plots showed fluctuating patch boundaries while heavily grazed plots showed a certain degree of patch stability. Conclusion: The model presented here reveals that selective grazing at the bite scale could lead to the creation of relatively stable patches within the pasture. Locally maintaining short cover heights would result in divergent within-plot vegetation dynamics, and thus favor the functional diversity of vegetation.