Résumé
Tropical forest ecosystems cover nearly half of the world's forest area. These ecosystems provide basic services for the well-being of our societies (e.g. regulation of material flows, soil preservation) are particularly affected (loss and degradation of forests) by global changes. New Caledonia, and its ultramafic substrates, is no exception to this trend. Mining and fires are responsible for forest regression, fragmentation, degradation and soil erosion. The archipelago is ranked among the hotspots of global biodiversity and the ecological restoration of these ecosystems is a major challenge. The objective of this thesis is to contribute to the ecological restoration of the forests of New Caledonia in order to limit the erosion of soils on ultramafic substrates and is part of the disciplinary field of the ecology of restoration. The scientific objective is to study the influence of mycorrhizal symbiosis and root traits on the erosion of Ferralsols developed on ultramafic substrates through a multi-scale approach. In the first part, we highlight the floristic and structural characteristics of plant communities on ultramafic substrates. These suggest an increased expression of the root traits of dominant species in monodominant forests. The central place of fire and cyclones in the dynamics of plant succession is also emphasized, as well as mycorrhizal symbioses in maintaining the monodominance of Nothofagus aequilateralis. In the second part, our work suggests the existence of a biotic threshold between degraded ligno-herbaceous maquis and ligno-herbaceous maquis. Repeated fires would be the main cause of crossing this threshold. The increase in plant biomass would be the key factor contributing to the stability of the aggregates between these two stages. Thus, fire protection combined with the promotion of plant successions through the use of potentially facilitating species such as Costularia arundinacea could be an efficient strategy for limiting erosion of Ferralsols. In addition, the influence of plant (e.g. floristic composition) and fungal communities (e.g. biomass) on the stability of aggregates is clearly highlighted. The third part aims to evaluate the ability of plant species, combined with mycorrhizal fungi, to increase the stability of aggregates, but also to identify root traits and mycorrhizal associations contributing to this increase. In this section, we demonstrate that the combination of high values for “Root Mass Density - Root Length Density – percentage of fine roots” is effective in increasing aggregate stability. Moreover, the efficacy of this combination of root traits is enhanced by mycorrhizal symbiosis. Thus, of the three plant species evaluated, Costularia arundinacea would be the most effective tool species for increasing the stability of aggregates. This work allowed us to highlight the influence of root traits and mycorrhizal symbiosis on the stability of soil aggregates on Ferralsols on ultramafic substrates, at the scale of plant communities or at the level of plant species. To our knowledge, these results are the first to be acquired on this type of soil. These results make it possible to formulate proposals for the ecological restoration of forest ecosystems in New Caledonia in order to limit soil erosion on ultramafic substrates, but also to sketch new avenues of research.