23 24 KEYWORDS: biofertilizer, nitrogen fixation, plant growth promoter, genome 25 sequencing, computational phenotyping 26 2ABSTRACT Previous studies have shown that the sugarcane microbiome harbors diverse 27 plant growth promoting (PGP) microorganisms, including nitrogen-fixing bacteria, and the 28 objective of this study was to design a genome-enabled approach to prioritize sugarcane 29 associated nitrogen-fixing bacteria according to their potential as biofertilizers. Using a 30 systematic high throughput approach, 22 pure cultures of nitrogen-fixing bacteria were isolated 31 and tested for diazotrophic potential by PCR amplification of nitrogenase (nifH) genes, common 32 molecular markers for nitrogen fixation capacity. Genome sequencing confirmed the presence 33 of intact nitrogenase nifH genes and operons in the genomes of 18 of the isolates. Isolate 34 genomes also encoded operons for phosphate solubilization, siderophore production operons, 35 and other PGP phenotypes. Klebsiella pneumoniae strains comprised 14 of the 22 nitrogen-36fixing isolates, and four others were members of closely related genera to Klebsiella. A 37 computational phenotyping approach was developed to rapidly screen for strains that have high 38 potential for nitrogen fixation and other PGP phenotypes while showing low risk for virulence 39 and antibiotic resistance. The majority of sugarcane isolates were below a genotypic and 40 phenotypic threshold, showing uniformly low predicted virulence and antibiotic resistance 41 compared to clinical isolates. Six prioritized strains were experimentally evaluated for PGP 42 phenotypes: nitrogen fixation, phosphate solubilization, and the production of siderophores, 43 gibberellic acid and indole acetic acid. Results from the biochemical assays were consistent 44 with the computational phenotype predictions for these isolates. Our results indicate that 45 computational phenotyping is a promising tool for the assessment of benefits and risks 46 associated with bacteria commonly detected in agricultural ecosystems. IMPORTANCE A genome-enabled approach was developed for the prioritization of 48 native bacterial isolates with the potential to serve as biofertilizers for sugarcane fields 49 in Colombia's Cauca Valley. The approach is based on computational phenotyping, 50 which entails predictions related to traits of interest based on bioinformatic analysis of 51 whole genome sequences. Bioinformatic predictions of the presence of plant growth 52 promoting traits were validated with experimental assays and more extensive genome 53 comparisons, thereby demonstrating the utility of computational phenotyping for 54 assessing the benefits and risks posed by bacterial isolates that can be used as 55 biofertilizers. The quantitative approach to computational phenotyping developed here 56 for the discovery of biofertilizers has the potential for use with a broad range of 57 applications in environmental and industrial microbiology, food safety, water quality, and 58 antibiotic resistance studies.
6...