The application of a filter-based assembly model to develop best practices for Pannonian sand grassland restoration.
The application of a filter-based assembly model is a promising concept to support ecological restoration. The filters illustrate dispersal, abiotic and biotic constraints that narrow down the species pool and determine the realized species composition. The aim of restoration interventions is the manipulation of filters to achieve a restoration target. We tested the applicability of a filter-based assembly model in the restoration of endemic Pannonian sandy grassland on old-fields. We report the results of a six-year experiment to identify key filters and find possible interactions. Treatments included the following: (i) seeding (dispersal filter), (ii) carbon amendment to lower nitrogen availability (abiotic filter) and (iii) mowing (biotic filter). Treatments were carried out in 1 m2 plots following the same Latin square design in three old-fields. Seeding was the most important treatment increasing species richness, vascular cover and enhancing target species composition. Mowing played a secondary role, acting primarily in interaction with the other treatments. Carbon amendment significantly reduced soil mineral nitrogen, but played a subordinate role in determining vegetation composition. Significant interactions were found between the biotic and dispersal and the biotic and abiotic filters determining primarily the structural characteristics of the vegetation in terms of vascular cover, moss cover, litter and bare ground. Regarding species composition, synergistic links between the dispersal and biotic filters for seeded species and long-lived forbs, and antagonistic interaction between the dispersal and abiotic filters for seeded species were found. Synthesis and applications. Based on the impact and interactions of filters uncovered in this experiment, we propose a scheme for Pannonian grassland restoration, which can be validated and also used for other oligotrophic systems. In the proposed conceptual model, the dispersal filter plays a key role in determining the outcome of restoration measures followed by the biotic filter if manipulated by mowing, while the abiotic filter altered by nitrogen immobilization proved to have a weak effect. Based on our results, targeting several filters in parallel would improve restoration outcome. We propose that filter interactions should form an integral part of filter-based assembly models and should be taken into account in restoration decisions.