From degradation to restoration: microbial shifts and carbon cycling in Brazilian drylands

• Grant number: 25/04487-9
• Support Opportunities: Scholarships abroad - Research Internship - Doctorate
• Start date: November 01, 2025
• End date: June 30, 2026
• Field of knowledge: Agronomical Sciences - Agronomy - Soil Science
• Principal Investigator: Maurício Roberto Cherubin
• Grantee: Antonio Yan Viana Lima
• Supervisor: Noah Fierer
• Host Institution: Escola Superior de Agricultura Luiz de Queiroz (ESALQ). Universidade de São Paulo (USP). Piracicaba , SP, Brazil
• Institution abroad: University of Colorado Boulder, United States
• Associated to the scholarship: 24/03722-1 - Soil health assessment in desertified drylands in Brazil, BP.DR

Abstract

The Brazilian semiarid region is highly vulnerable to desertification, affecting over 34 million people and rendering extensive areas unfit for agriculture. The Caatinga biome has experienced significant degradation, losing 14.4% of its native vegetation between 1985 and 2023 due to overgrazing and agricultural expansion. These changes have accelerated soil erosion, reduced soil organic matter, and disrupted ecosystem functions, including carbon cycling. Restoration practices have shown the potential to reverse these effects by enhancing soil microbial diversity and functionality. This study exploresmicrobial diversity and metabolic functions in soils under three land-use scenarios - native vegetation, desertified and restored areas - within Brazil's Desertification Nuclei. We hypothesize that desertification not only reduces microbial diversity but also alters the composition and functionality of microbialcommunities, leading to disruptions in key processes such as the C cycle. Soil samples were collected (0 to 10 cm depth) from four desertification nuclei (Irauçuba-CE, Cabrobó-PE, Gilbués-PI, and Cariris Velhos-PB), representing different land-use scenarios. Shotgun metagenomic sequencing has already been completed. The bioinformatics workflow includes quality control (FastQC, Trimmomatic) genome assembly (MEGAHIT), taxonomic classification (Kraken2, MetaPhlAn), and functional annotation (KEGG, EggNOG, Pfam). Multivariate analyses (PCA, RDA) and co-occurrence network analysis (CoNet, Cytoscape) will be employed to investigate microbial interactions and their relationship with soil health indicators. Additionally, machine learning models (e.g., Random Forest) will be used to predict soilfunctionality and identify key microbial and environmental factors driving ecosystem restoration. These analyses will provide a deeper understanding of how microbial communities, and their functions respond to degradation and restoration, offering valuable insights into the potential for ecological restoration tomitigate the impacts of desertification in the Brazilian semiarid region. (AU)