Banca de DEFESA: THALLYTA DAS GRACAS ESPINDOLA DA SILVA

Exploring microbial diversity: bioprospecting diazotrophic bacteria in rhizospheres of productive pastures

Biological nitrogen fixation; Sustainability; Soil quality; Machine learning

The increasing environmental degradation caused by the intensive use of agricultural inputs has driven the search for sustainable practices. Among these, the use of bioinputs, such as diazotrophic bacteria, stands out. These bacteria are capable of fixing atmospheric nitrogen and contributing to the reduction of synthetic fertilizer use. This study aimed to: (i) compare the recruitment capacity of diazotrophic bacteria by Brachiaria decumbens and Digitaria eriantha; and (ii) evaluate the effects of applying bacterial isolates, combined or not with grape pruning biochar, on the development of Brachiaria brizantha and soil quality. In the first experiment, rhizosphere soil samples were collected from natural pastures in the Agreste region of Pernambuco. The bacteria were isolated on JNFb medium, purified, and characterized by Gram staining. To measure microbial recruitment, a Composite Recruitment Index (CRI) was used, which integrated precocity, dilution diversity, and detection rate. B. decumbens exhibited greater recruitment of diazotrophic bacteria, with 81.6% positive detections and an average ICR of 0.775, compared to 69.4% and 0.706 for D. eriantha, although there was no statistically significant difference. In the second experiment, five bacterial isolates were used to inoculate B. brizantha seeds, with or without biochar. The analyses showed that biochar had a significant effect on 66.7% of the variables, such as plant height, biomass, and soil pH. Bacterial isolates influenced 26.7% of the variables, with isolate bac21 being the most promising. Significant interactions between isolate and biochar were also observed. The treatments increased urease activity and soil phosphorus and nitrogen contents, suggesting a biofertilizing effect. Principal component analysis (PCA) revealed a clear separation between treatments, primarily due to the isolates. Machine learning techniques, such as LDA and Random Forest, indicated that dry leaf weight and soil pH were the most relevant variables in distinguishing between groups, although the small number of samples limited statistical robustness. It is concluded that B. decumbens has a greater potential for microbial recruitment, and that the combination of bacterial isolates with biochar favors the initial growth of B. brizantha and improves soil chemical properties. These results reinforce the potential of the integrated use of microorganisms and biochar as a viable strategy to promote sustainability in forage production