On the importance of fine-scale spatial complexity in vegetation restoration studies

Effective restoration should start from an understanding of the spontaneous processes of vegetation succession and utilize the natural 'self-repair' mechanisms. The number of possible restoration treatments (for example, manipulating the level of soil nutrients, propagule sources or disturbance regime) is limited. However, the relatively constant treatments should interact with a great variability of ecosystem states and landscape contexts, making restoration practice a very challenging and hard task. This context dependence of local vegetation dynamics is emphasized by the non-equilibrium ecological paradigm. This paradigm views the developing plant community as a complex dissipative system, which involve a methodology with explicit representation of spatiotemporal patterns. Restoration practice needs simple methods that easy to implement in the routine. However, there is a conflict between the simplicity required by the application and the complexity offered by advanced theory. We propose a solution based on the information theory models of Juhász-Nagy. These models are able to represent complex community patterns in a very simple way. The frequency distribution of species combinations within the communityis detected as a function of spatial resolution. Comparing the pattern of species combinations detected in the field with other reference patterns generated by neutral models, we are able to quantify and interpret constraints of vegetation dynamics in an explicit, detailed way. The related information theory models are additive that makes the calculations easy. The basic models are very simple and practical in routine works. Nevertheless the more advancedmodels of the model family can be connected with spatially explicit individual based models and with advanced techniques analysing complex trajectories in abstract coenostate spaces. The pattern of species combinations can be sampled with long transects in the field. This sampling is rapid and causes minimum samp