Agromining aims to rehabilitate natural metal-rich soils (ultramafic soils) by extracting metals of high economic importance, such as nickel (Ni), using hyperaccumulator plants and then to recover these metals for industrial purposes. Ultramafic soils are characterized by low fertility levels and this can limit yields of hyperaccumulators and metal. Here, we characterized the potential benefits for phytoextraction efficiency of co-cropping two plants: a Ni-hyperaccumulator (Odontarrhena chalcidica; Brassicaceae) and a legume (Vicia sativa; Fabaceae). A two years field experiment was set up in an ultramafic zone (North West Spain). Three treatments were tested: co-cropping (CoC), fertilized control with ammonitrate of the hyperaccumulator (FCon) and non-fertilized control (NFCon). Over the two years, FCon and CoC treatments had significantly higher yield values (plant biomass and Ni) compared to NFCon. Indeed, after two years of cultivation, CoC had increased O. chalcidica’s biomass by 24% compared to FCon treatment. Moreover, CoC had higher Ni-yields than FCon for the first (7.8 vs 5.8 kg ha^-1) and the second year (13.6 vs 9.9 kg ha^–1). A Principal Component Analysis (PCA), based on the microbiological and physico-chemical soil parameters measured each year, showed that all treatments of the second year cultivation were negatively correlated with soil bioavailable Ni concentration (Ni-DTPA) and phosphatase microbial activity. Moreover, FCon was characterized by negative correlations with almost all microbial and physico-chemical soil parameters. A non-metric multidimensional scaling analysis (NMDS) of the operational taxonomic units (OTUs) showed that soil bacterial diversity of the first year treatments was clearly separated from that of the second year. Moreover, after two years of cultivation, CoC increased the relative abundance of the phylum Gemmatimonadetes when compared to other treatments, while relative abundances of γ-Proteobacteria were significantly reduced over time for all the treatments. Multivariate regression tree (MRT) provided a tree with three terminal nodes based on exchangeable Ni (Niex) and soil organic carbon (Corg). Niex was the most influencing parameter on the phyla’s relative abundance with all samples collected during the first year characterized by low Niex levels (< 1.18 mg kg^–1) and separated from those of the second year with higher Niex. Concerning samples from the first year, Corg was identified as the second major environmental factor and separated CoC (Corg > 3.18 g kg^–1) from others treatments (Corg < 3.18 g kg^–1). Based on the predicted metagenomes, genes belonging to different families were identified depending on the treatments. This study showed that co-cropping a hyperaccumulator with a legume in agronomic systems dedicated to Ni-agromining improved plant biomass production and Ni-yields as in the case of fertilized control, but co-cropping less impacted some microbial soil parameters. Ameliorating agromining by replacing mineral fertilizers would combine eco-efficient or sustainable metal recovery.