Search

Understanding the effects of forest-to-agriculture conversion on microbial diversity has been a major goal in soil ecological studies. However, linking community assembly to the ruling ecological processes at local and regional scales remains challenging. Here, we evaluated bacterial community assembly patterns and the ecological processes governing niche specialization in a gradient of geography, seasonality, and land-use change, totalizing 324 soil samples, 43 habitat characteristics (abiotic factors), and 16 metabolic and co-occurrence patterns (biotic factors), in the Brazilian Atlantic Rainforest, a subtropical biome recognized as one the world’s largest and most threatened hotspots of biodiversity. We observed no significant shifts in alpha diversity due to land-use change. Meanwhile, pairwise beta diversities and distance decays were lower in pastures than those observed for forest and no-till soils. Pasture communities showed a predominantly neutral model, regarding stochastic processes, with moderate dispersion, leading to biotic homogenization. Most no-till and forest microbial communities followed a niche-based model, with low rates of dispersal and weak homogenizing selection, indicating niche specialization or variable selection. Historical and evolutionary contingencies, as represented by soil type, season, and dispersal process were the main drivers of microbial assembly and processes at the local scale, markedly correlated with the occurrence of endemic microbes. Our results indicate that the patterns of assembly and their governing processes are dependent on the niche occupancy of the taxa evaluated (generalists or specialists). They are also more correlated with historical/evolutionary contingencies and the interactions among taxa (i.e. co-occurrence patterns) than the land-use change itself. Keywords: Biodiversity hotspots, historical contingency, land-use change, microbial distance-decay, microbial niche specialization, soil bacterial co-occurrence.

Forest-to-agriculture conversion has been identified as a major threat to soil biodiversity and soil processes resilience, although the consequences of long-term land use change to microbial community assembly and ecological processes have been often neglected. Here, we combined metagenomic approach with a large environmental dataset, to (i) identify the microbial assembly patterns and, (ii) to evaluate the ecological processes governing microbial assembly, in bulk soil and soybean rhizosphere, along a long-term forest-to-agriculture conversion chronosequence, in Eastern Amazon. We hypothesized that (i) microbial communities in bulk soil and rhizosphere have different assembly patterns and (ii) the weight of the four ecological processes governing assembly differs between bulk soil and rhizosphere and along the chronosequence in the same fraction. Community assembly in bulk soil fitted most the zero-sum multinomial (ZSM) neutral-based model, regardless of time. Low to intermediate dispersal was observed. Decreasing influence of abiotic factors was counterbalanced by increasing influence of biotic factors, as the chronosequence advanced. Undominated ecological processes of dispersal limitation and variable selection governing community assembly were observed in this soil fraction. For soybean rhizosphere, community assembly fitted most the lognormal niche-based model in all chronosequence areas. High dispersal and an increasing influence of abiotic factors coupled with a decreasing influence of biotic factors were found along the chronosequence. Thus, we found a dominant role of dispersal process governing microbial assembly with a secondary effect of homogeneous selection process, mainly driven by decreasing aluminum and increased cations saturation in soil solution, due to long-term no-till cropping. Together, our results indicate that long-term no-till lead community abundances in bulk soil to be in a transient and conditional state, while for soybean rhizosphere, community abundances reach a periodic and permanent distribution state. Dominant dispersal process in rhizosphere, coupled with homogeneous selection, brings evidences that soybean root system selects microbial taxa via trade-offs in order to keep functional resilience of soil processes.

The conversion of native forest to agriculture is the main cause of microbial biodiversity loss in Amazon soils. In order to better understand this effect, we used metagenomics to investigate microbial patterns and functions in bulk soil and rhizosphere of soybean, in a long-term forest-to-agriculture conversion. Long-term forest-to-agriculture led to microbial homogenization and loss of diversity in both bulk soil and rhizosphere, mainly driven by decreasing aluminum concentration and increased cations saturation in soil, due to liming and fertilization in long-term no-till cropping. Data revealed that long-term no-till cropping culminated in a decrease in Acidobacteria, Actinobacteria and Proteobacteria abundances. However, α- and β-Proteobacteria abundances were higher in the rhizosphere than in bulk soil, regardless of the time after forest-to-agriculture conversion. Changes in functional potential occurred predominantly in bulk soil, with decreases in functions related to potassium metabolism and virulence, disease and defense, while functions related to nucleic acids metabolism increased. Functions in the soybean rhizosphere remained stable, except for those related to potassium metabolism, which decreased after 20-year no-till cropping. Together, our results show that the soybean root system selects microbial taxa via trade-offs, to maintain functional resilience in the rhizosphere microbiome over time.

We evaluated the bacterial and archaeal community dynamics and assembly in soils under forest, grassland and no-till cropping, using a high-throughput shotgun metagenomics approach. No significant alterations in alpha diversity were observed among different land uses, but beta diversity in grassland was lower than that observed in forest and no-till soils. Grassland communities showed assembly that predominantly followed the neutral model, i.e. high homogenizing selection with moderate dispersion, leading to biotic homogenization. Both no-till and forest soil communities were found to have assembly that predominantly followed a niche model, i.e. low rates of dispersal and weak homogenizing selection, resulting in maintenance of higher beta diversity relative to grasslands, indicating niche specialization or variable selection. Taken together, our results indicate that the patterns of assembly and their governing processes are dependent on the land use employed after deforestation, with consequences for taxa turnover and microbial functional potential.