Forest tree genetic resources are one of the most important repositories of biological diversity, constituting a key component for the stability of the ecosystems. Due to the increasing human pressure and global climate changes, the management and conservation of forests has become a key issue in order to sustain their productive, environmental and socio-economic value. Therefore, worldwide the development of appropriated conservation strategies became a priority and several international, regional and national initiatives and framework instruments have been established. The research focused on the analysis of plant and microbial diversity in the Miombo and Mopane woodlands, the densely-forested ecosystems of African savannas, in two key conservation areas, the Niassa Special Reserve (NSR), northern Mozambique, and Limpopo National Park (LNP), southern of Mozambique, respectively. Both ecosystems are considered sources of woody species, their resources contributing to the livelihood systems of millions of households in rural and urban areas that depend on these ecosystems for their food, health, energy and housing needs. Moreover, these ecosystems play an important role for the environment, for its overall impact on energy, water and carbon balance. Fire is the major threat to which these ecosystems are exposed, being essentially related to human and climate pressure and herbivory by elephants. The project aimed at contributing to the Management Program of NSR and LNP, particularly to the components of fire control and biodiversity conservation. The proposed work was integrated into a multi-disciplinary program and focused on two selected priority species from the legume family, Brachystegia boehmii (Miombo) and Colophospherum mopane (Mopane), preceeded by an overall characterization of legume tree distribution, diversity and conservation status in the two ecosystems. In the first part of the study it was concluded that, despite the high diversity of species (78), there was a notable research gap regarding their conservation status (Chapter 2). Over the last two years, this gap has been properly addressed and the percentage of catalogued species in the International Union for Conservation of Nature’s Red List of Threatened Species – IUCN, has increased from 15 to 77%. Most of the species (56) are classified as least concern, 2 as vulnerable, and another 2 as nearly threatened, reflecting the ongoing conservation efforts in the region. The remaining 18, in which the two focal species of this thesis are included, were not yet evaluated (Chapter 5). In the second part of the the work (Chapters 3 and 4), the dynamics of rhizobacteria diversity and associated functional traits have been analysed, considering different soil types and fire regimes, using culture-independent (Metagenomics) and culture dependent approaches. Despite the fact that both Miombo and Mopane rhizosphere bacteria were quite rich, there was a clear difference between them. First, the abundance of Operational Taxonomic Units (OTUs) was higher in B. boehmii (977 – 1391 OTUs) than in C. mopane (696 – 831 OTUs), which was also reflected in the Shannon diversity index, i.e. 7.21 – 8.50 vs 4.040 – 4.087, respectively. Second, while the bacterial diversity in the rhizosphere of B. boehmii varied with the soil type and fire frequency, no changes were observed in C. mopane. This may be attributed to the multi-specific nature of Miombo vs. the mono-specific nature of Mopane. However, in both cases, a pyrodiversity effect seems to be an important evolutionary driver towards high-temperature tolerance in both cases. The most representative groups were Actinobacteria (40%), Proteobacteria (20%), and Acidobacteria (13%) in Miombo and Firmicutes (31%), Bacteroidetes (25%), and Proteobacteria (23%). At the species level, the rhizosphere of B. boehmii was enriched in newly characterized species from recalcitrant environments, like the thermophilic, halo- and radiotolerant Gaiella oculta and the thermotolerant Rhodoplanes tepidicaeni and Saccharopolyspora spongiae, likely playing multiple roles in plant growth and protection as well as in soil fertility and/or restauration and/or remediation. The metagenomics approach did not allow to resolve taxonomic questions at the species level in the Mopane rhizosphere, probably because the most abundant species are yet unidentified. However, higher taxonomic classification suggests that the diversity of functions accomplished by these bacteria is similar to that of Miombo. Regarding the Plant Growth-Promoting Bacteria (PGPB) isolated from soils, a larger collection was captured in Mopane (16 isolates, nine species: Bacillus sp., Caballeronia concitans, Pantoea agglomerans, Paraburkholderia phenoliruptrix, Phyllobacterium myrsinacearum, Pseudomonas azotoformans, Pseudomonas gessardii, Pseudomonas synxantha and Stenotrophomonas maltophilia) than in Miombo (eight isolates, five species: Burkholderia sp., Caballeronia zhejiangensis, Microvirga sp., Rhizobium spp., Variovorax defluvii), all accomplishing at least one PGP trait, i.e. nitrogen fixation, phosphate solubilization, indole acetic acid production, siderophore production, and hydrolysis of cellulose. Regarding the bacteria trapped in Vigna unguiculata (cowpea) nodules, the number of isolates was considerably higher in Miombo (64 isolates, including symbiotic Ensifer adhaerens, Mesorhizobium sp., Neorhizobium galegae, Rhizobium sp., and non-symbiotic Agrobacterium sp., Cohnella sp., Herbaspirillum huttiense, Pseudomonas sp., and Stenotrophomonas sp.) than in Mopane (23 isolates, including symbiotic Bradyrhizobium sp. and Rhizobium sp., and non-symbiotic Azospirillum zeae, Cohnella rhizosphaerae, P. agglomerans, Pseudomonas nitroreducens. In both cases, nodule consortia bacteria promoted cowpea growth, but the effect of Miombo consortia was considerably more prominent. Altogether the results revealed that rhizospere bacteria from both Miombo and Mopane are highly resilient to soil composition and fire, hiddding an enourmous bio-economic potential for the devolpement of nature-based solutions like, the production of fertilizers, soil remediators, phytostimulators, or antimicrobials. The produced knowledge constitutes a significant contribution to the understanding of the dynamics of soil bacteria in tropical forest ecosystems, namely in what concerns plant-microbe interactions, and bioprospection of new species. Despite the fact that tropical ecosystems host ca. 80% of the global biodiversity, pressures imposed by a combination of human, animal and climate pressure are imposing major challenges. Thus, understanding key aspects of the ecosystem dynamics is of utmost importance to support community-based conservation programs.