Biogeography of Soil Bacterial Networks along a Gradient of Cropping Intensity

Battle Karimi,Sébastien Terrat,Corinne Cruaud,Lionel Ranjard,Antonio Bispo,Nicolas Chemidlin Prévost-Bouré,Patrick Wincker,Dominique Arrouays,Samuel Dequiedt,Claudy Jolivet

doi:10.1038/s41598-019-40422-y

Abstract

Although land use drives soil bacterial diversity and community structure, little information about the bacterial interaction networks is available. Here, we investigated bacterial co-occurrence networks in soils under different types of land use (forests, grasslands, crops and vineyards) by sampling 1798 sites in the French Soil Quality Monitoring Network covering all of France. An increase in bacterial richness was observed from forests to vineyards, whereas network complexity respectively decreased from 16,430 links to 2,046. However, the ratio of positive to negative links within the bacterial networks ranged from 2.9 in forests to 5.5 in vineyards. Networks structure was centered on the most connected genera (called hub), which belonged to Bacteroidetes in forest and grassland soils, but to Actinobacteria in vineyard soils. Overall, our study revealed that soil perturbation due to intensive cropping reduces strongly the complexity of bacterial network although the richness is increased. Moreover, the hub genera within the bacterial community shifted from copiotrophic taxa in forest soils to more oligotrophic taxa in agricultural soils.

Highlights

Analyzing the microbial co-occurrence network in soil could provide a promising way to improve our understanding of soil microbial community regulation, functioning and stability[10,12]
Soil bacterial co-occurrence networks have been shown to be impacted by soil physico-chemical characteristics[16,17] and climate[18], to vary with age of the plant cover[15] and to change with the soil land use[19,20]. Both studies by Morriën and Lupatini demonstrated that the least complex bacterial networks were found in cropped soils, i.e. those most disturbed by agricultural practices, as compared to pastured or forest soils
The connections were the significant correlations between the nodes, which corresponded to the genera occurring in the soils under the respective land use

Summary

Introduction

Analyzing the microbial co-occurrence network in soil could provide a promising way to improve our understanding of soil microbial community regulation, functioning and stability[10,12]. Soil bacterial co-occurrence networks have been shown to be impacted by soil physico-chemical characteristics[16,17] and climate[18], to vary with age of the plant cover[15] and to change with the soil land use[19,20] Both studies by Morriën and Lupatini demonstrated that the least complex bacterial networks were found in cropped soils, i.e. those most disturbed by agricultural practices, as compared to pastured or forest soils. Previous studies based on RMQS data had revealed a lower microbial biomass but a higher bacterial taxonomic richness in vineyards and crop system than in forest and grassland soils[3,22] Since these two microbial parameters were differently impacted by the land uses, investigating other microbial community parameters, such as the co-occurrence network, is a major concern to understand more comprehensively the response of microbial communities to the soil perturbations. We identified the hub taxa for the bacterial network in each land use and related the ecological attributes of these taxa to the environmental and management context

Methods

Results

Discussion

Conclusion