The response of soil multi-functionality to agricultural management practices can be predicted by key soil abiotic and biotic properties

Abstract

Abstract Soil organisms play an essential role in ecosystem functioning, including nutrient cycling and climate regulation. Most of soil functions have a microbial origin and recent findings suggest that various microbial community attributes (e.g. diversity, composition, abundance) are linked to the rate of soil functions in natural ecosystems. However, no such evidence is available from the agronomically managed systems. The types of agronomic practice and mechanisms (biotic vs. abiotic) that could impact microbial regulation of soil functions remain poorly understood. We collected soil samples from three long-term field trials across three geographically distinct regions with multiple management treatments (e.g. tillage, stubble management and fertiliser application) and measured the response of a number soil abiotic (e.g. soil aggregate stability, pH, nutrient status) and biotic (e.g. bacterial and fungal community structure, abundance of various microbial taxa) variables, and soil functions linked to nutrient cycling and nutrient availability. We used information theory and multi-model inference to identify key biotic and abiotic predictors of soil functions, and to model their response to land management through a multi-functionality index. Results indicated that no-till treatment generally impacted microbial community (42% decrease in the average gene copy number of all microbes compared to other management practices) and promoted soil structure (27% increase in the small aggregate fraction) but reduced some process rates, while stubble retention increased nutrient availability in the presence of fertilisation (59% on average). But these responses were site specific. The multi-functionality index associated with nutrient cycling declined under no-till system (47% on average) but increased when no-till was combined with stubble retention (49% on average) and fertiliser application (44%). Our modelling suggested that the best models that explained most of the variation in MF always included both abiotic and biotic variables. Amongst the latter, fungal community structure and beta(β)-Proteobacteria abundance were of importance as the model’s R² dropped by 12% when biotic attributes were removed providing evidence that the link between microbial community and soil functions was maintained in agricultural soils. This study provides support for the inclusion of microbial data in decision support systems for sustainable farming practices.