Ultramafic soils enriched in nickel, such as found in Australia and New Caledonia, are associated with unique, diverse and poorly known vegetation communities. Re-establishment of these highly specific ecosystems is still a challenge for Ni mining companies. Ultramafic vegetation communities are the outcome of a long evolution process resulting in their adaptation to the extreme soil conditions found on ultramafic outcrops. Mycorrhizal fungi, a very common plant symbiont, are generally thought to be beneficial to plants in other ecosystems, providing plants with phosphorus and even promoting metal tolerance in plants in some cases. We examined the hypothesis that mycorrhizal fungi may contribute to the survival of plants in ultramafic soil conditions. Bandalup Hill, an ultramafic outcrop enriched in Ni (South West of Western Australia) was selected to assess the contribution of mycorrhizal fungi to ultramafic plants. Soil constraints, in particular the degree of Ni toxicity, were assessed at two sites with ultramafic soils within the outcrop. Total metal, nutrient, DTPA extractable Ni and available P were measured in soil while Ni, Ca and Mg were tested in the soil solution. In addition, nutrients and metals were analyzed in shoots of some plant species occurring at each site: Eucalyptus flocktoniae, Melaleuca pomphostoma, Melaleuca coronicarpa and Hakea verucosa. Topsoils in Bandalup Hill and plant shoots had high levels of Ni, and very low levels of P, K and N. Variation in DTPA extractable Ni between sites reflected the variation in shoot Ni level of E. flocktoniae and M. pomphostoma. Variations in soil solution Ni levels reflected variations in shoot Ni levels of M. coronicarpa and H. verucosa between sites. The germination requirements of the plant species used to assess the soil constraints was assessed. Species selected included Eucalyptus flocktoniae, Melaleuca coronicarpa, and Hakea verucosa. Seeds of E. flocktoniae and M. coronicarpa had a higher germination rate if pre-treated with smoke water, while no pre-treatment was required to germinate H. verucosa seeds. The unusual germination requirement of E. flocktoniae and M. coronicarpa involve complex chemical signals that may be present in the soil when the conditions are more favorable for plant establishment. Such unusual germination requirement may represent an adaptation to the hostile conditions of the ultramafic soils of Bandalup Hill. The mycorrhizal association and root characteristics of the selected plant species was also assessed after 8 weeks of growth in undisturbed ultramafic topsoil cores from Bandalup Hill. Roots of these species (including H. verucosa from a previously designated non-mycorrhizal family, Proteaceae) were associated with mycorrhizal fungi. Roots of E. flocktoniae and M. coronicarpa were colonized by both arbuscular mycorrhizal fungi (AMF) and ectomycorrhizal fungi (ECM), while roots of H. verucosa only contained some AM fungal structures. All species had high shoot to root ratios and their root characteristics reflected their association with mycorrhizal fungi. Based on the previous observations, uninoculated and inoculated E. flocktoniae seedlings were grown for 10 to 16 weeks in sand amended with Ni at 0, 0.2, 1 and 2.3 mg/kg. Mycorrhizal inoculum consisted of spores of Pisolithus sp. (ECM) or a mix of AMF spores and colonized root fragments, both originating from Bandalup Hill. Another inoculum consisted in Pisolithus sp. spores from a site with ultramafic soils in New Caledonia. Inoculation with AM and ECM fungi from Bandalup Hill was beneficial to E. flocktoniae. Benefits consisted mainly of a reduction of Ni shoot translocation at the highest Ni soil level. At 1 mg/kg soil Ni, E. flocktoniae exhibited a certain degree of tolerance to Ni. A substantial increase in growth and nutrient uptake with Pisolithus sp. from Western Australia was also observed. The contribution of AM fungi from Bandalup Hill to E. flocktoniae, M. coronicarpa, H. verucosa, and Trifolium subterraneum (clover) was then examined in ultramafic soil from Bandalup Hill.Steaming of ultramafic soil increased the availability and plant uptake of P. Consequently, uninoculated seedlings grew better, and inoculation with AM fungi decreased the growth of native plant species but did not affect their shoot Ni concentration. The presence of AM fungi increased the concentration of P in shoots of native plants species. Inoculation had no effect on the growth and nutrient content of subterranean clover. As mining activities have the potential to reduce the infectivity of AM fungi in topsoils, the effect of disturbance and storage practices on the AM infectivity of ultramafic topsoils collected in summer or winter from Bandalup Hill was investigated. Disturbance consisted in passing topsoil through a 2mm seive and cutting roots into 1cm fragments. Disturbed topsoil was then stored at room temperature in pots that were either sealed from the atmosphere or left open, and pots were maintained at field capacity. E. flocktoniae seedlings were planted in undisturbed and disturbed topsoil just after topsoil collect and then after 3, 6 and 9 months of topsoil storage. AM fungi present in the topsoil collected in summer was less susceptible to initial disturbance than AM fungi present in topsoil collected during winter. Also, storage of topsoil in sealed pots watered to field capacity was more detrimental to its infectivity than storage of topsoil in dry conditions. Mycorrhizal fungi can contribute to the survival of some native plant species in the ultramafic soils of Bandalup Hill and they may represent another strategy to improve the success of Ni mine revegetation. However, such contribution may not be the unique avenue for native plants to survive in ultramafic soils of Bandalup Hill.

La connaissance de la diversité microbienne des milieux ultramafiques est essentielle pour établir le fonctionnement écologique de ces milieux, qui présentent de fortes teneurs en Ni et sont caractérisés par une flore particulière, e.g. plantes hyperaccumulatrices de Ni. La rhizosphère des hyperaccumulateurs comporte une forte proportion de bactéries résistantes au Ni, qui peuvent aussi agir sur la nutrition des plantes et sur les propriétés physico-chimiques du sol. Le premier défi de cette thèse a été de cerner le déterminisme de la diversité bactérienne de la rhizosphère d’hyperaccumulateurs de Ni. Le second a été de tester l'intérêt de souches PGPR (Plant Growth Promoting Rhizobacteria) pour optimiser l'agromine à partir d'interactions entre les rhizobactéries et les hyperaccumulateurs de Ni. La démarche s'est appuyée sur un ensemble de prospections dans deux régions climatiques et sur des analyses de séquençage haut débit. Des tests de cultures de plantes hyperaccumulatrices inoculées ont également été conduits. Les résultats montrent que le déterminisme de la diversité bactérienne est variable selon l'échelle spatiale. A l'échelle mondiale, le type de végétation est le facteur majeur structurant les communautés bactériennes, elle-même contrôlée indirectement par le climat. L’influence directe du climat (température et humidité) sur la diversité est significative mais moindre. A l'échelle d'une région climatique, la physico-chimie des sols ultramafiques structure et détermine la diversité des communautés bactériennes rhizosphériques. Enfin, l'inoculation de souches PGPR fortement bioaccumulatrices de Ni modifie la dynamique du Ni dans le sol, ce qui démontre qu'il existe une compétition pour le Ni entre la plante et la bactérie inoculée. En conclusion, le déterminisme de la diversité des communautés bactériennes rhizosphériques est dépendant de l'échelle spatiale considérée. En outre, le choix de la souche PGPR à inoculer, dans un contexte d'amélioration de l'agromine du Ni, est primordial. / Knowledge of the microbial diversity in ultramafic areas is essential to establish the ecological functioning of these environments, which display high level of Ni and are characterized by the presence of particular plants, e.g. Ni hyperaccumulators. The rhizosphere of these plants promotes a high proportion of Ni resistant bacteria that can act on plant nutrition and soil physicochemical properties. The first challenge of this thesis was to understand the bacterial rhizosphere diversity of Ni hyperaccumulators. The second was to test the interest of PGPR (Plant Growth Promoting Rhizobacteria) strains in order to improve agromining based on rhizobacteria and Ni hyperaccumulators interactions. The approach was based on two-contrasted climatic areas prospection and on high-throughput sequencing analyzes. Tests on culture of hyperaccumulator plants inoculated were also conducted. The results show that the determinism of this bacterial diversity is variable according to the spatial scale. On a global scale, the vegetation type, indirectly influenced by the climate, is the major factor structuring bacterial communities. The direct influence of the climate (temperature and humidity) on bacterial diversity is significant but lower. At the scale of a climatic region, the physic-chemistry of ultramafic soils structures and determines the rhizosphere bacterial community diversity. Finally, the inoculation of highly Ni bioaccumulative PGPR strains modifies the Ni dynamic in the soil, demonstrating that there is a competition for this metal between the inoculated bacteria and the hyperaccumulator plant. In conclusion, the rhizosphere bacterial community diversity is dependent on the considered spatial scale. Furthermore, these results emphasize how the choice of the PGPR strain to inoculate is important in order to improve Ni agromining.