Soil microbiome indicators can predict crop growth response to large-scale inoculation with arbuscular mycorrhizal fungi

Agricultural intensification, while boosting yields, has caused significant biodiversity loss, soil degradation, pollution, greenhouse gas emissions, and water eutrophication. The need for sustainable agricultural practices is paramount to ensure food security for a growing population while mitigating environmental damage. Soil ecological engineering, focusing on promoting beneficial soil biota, is a key strategy. Arbuscular mycorrhizal fungi (AMF), forming symbiotic relationships with most terrestrial plants, offer substantial potential. AMF enhance plant nutrient uptake (especially phosphorus), improve soil structure, enhance nutrient retention, reduce greenhouse gas emissions, improve drought tolerance, and bolster disease resistance. Two approaches exist to harness AMF benefits: promoting native AMF communities through sustainable practices or introducing AMF through inoculation. While greenhouse studies often show positive effects, field inoculation results are highly variable, making their widespread application unreliable. To improve the predictability and economic viability of AMF inoculation, this study aimed to identify key predictors of mycorrhizal growth response (MGR) in field settings, paving the way for a reliable soil diagnostic tool for farmers.

Numerous studies have demonstrated the positive effects of AMF inoculation on plant growth in greenhouse and, to a lesser extent, field settings. However, a meta-analysis of plant responses to mycorrhizal fungal inoculation revealed significant context dependency, with effects ranging from beneficial to detrimental. The inconsistency in field inoculation success is partly due to the unknown extent of introduced AMF establishment. Previous research highlighted the importance of AMF for plant nutrition, but lacked the comprehensive use of soil characteristics and molecular-based soil microbiome analysis to predict successful AMF application. Studies have also indicated a potential negative association between phosphorus availability and inoculation success, although this relationship isn't universally consistent.

Inoculation trials with the native AM fungus *Rhizoglomus irregulare* SAF22 were conducted over three years (2018-2020) across 54 maize fields in Northern Switzerland. The study employed a split-plot design in 2018 and a randomized complete block design in 2019 and 2020. Fields received standard N and K fertilization, with a subset in 2018 receiving additional P fertilization. A total of 52 soil parameters were measured before inoculation, encompassing chemical, physical, and biological properties. Soil and root fungal communities were analyzed using long-read sequencing (PacBio SMRT technology), targeting the ITS region. The mycorrhizal growth response (MGR) was calculated as the percentage change in maize biomass in inoculated plots relative to control plots. Statistical analyses included pairwise correlations, multivariate models (random forest, stepwise model selection, and exhaustive screening with glmulti), indicator species analysis, differential abundance analysis (DESeq2), and linear regression to identify key predictors of MGR. Root colonization by AMF was assessed using microscopy. The study employed a binary classification approach to assess model accuracy in predicting inoculation success (positive growth response >12.2% vs. neutral or negative response).

MGR varied significantly across fields, ranging from -12% to +40%. Significant positive growth responses (12-40%) were observed in 14 of the 54 fields (26%). Multivariate analyses revealed that soil fungal microbiome indicators were more important (53% of variance explained) in predicting MGR than soil parameters (29%). A model combining 15 soil parameters and 13 soil OTUs explained 86% of the variation in MGR (P < 0.001). A simplified model using only 3 high-MGR OTUs, 3 low-MGR OTUs, and four soil parameters still explained 68% of the variance. The abundance of pathogenic fungi in the soil was the strongest predictor of AMF inoculation success. The genus *Trichosporon*, a known human pathogen but not previously linked to plant pathogenicity, emerged as the most important predictor of high MGR, although its presence was low in high-MGR fields where AMF inoculation positively impacted yield. Root microbiome analysis showed that, in high-MGR fields, the introduced AMF suppressed the relative abundance of several plant pathogenic taxa including *Olpidium*, *Cladosporium*, *Mycochaetophora*, *Pyrenochaeta*, and *Vishniacozyma*. There was no correlation between root colonization and plant growth response; instead, the functional differences explained the MGR variance. Cross-validation revealed high mean accuracy (80-83%) in predicting whether inoculation would provide a significant benefit.

This study demonstrates that field inoculation with AMF can significantly increase maize yield, with some fields showing considerably higher increases than those achieved annually through breeding or cover crops. The findings highlight the importance of soil pathogen suppression by AMF. The study's success in predicting MGR based on soil microbial indicators contrasts with previous research, which emphasized nutrient availability. The weak influence of phosphorus, despite its substantial variation in the study fields, suggests that other factors may dominate in soils already above the phosphorus deficiency threshold. The association of positive growth responses with lower soil organic carbon and microbial biomass carbon suggests AMF are particularly beneficial in less healthy soils. The identification of *Trichosporon* as a key predictor requires further investigation to confirm its plant pathogenicity. The observed suppression of pathogenic fungi in inoculated high-MGR fields roots supports the hypothesis that AMF can contribute to pathobiome management.

This study provides a significant advancement in predicting the success of AMF field inoculations. A simple and cost-effective diagnostic tool using soil microbial indicators is feasible, and the approach can be extended to other biofertilizers. Future research should focus on testing different maize varieties and expanding the study's scope to diverse soil types, climatic zones, and agricultural systems with reduced agrochemical use.

The study focused on a single maize variety and geographic area. The results may not be directly generalizable to other maize varieties or regions. The use of AMF-specific primers for root microbiome analysis may have limited the detection of certain fungal taxa, including some pathogens identified in the soil data. While the model demonstrates high accuracy, further research is needed to validate its performance across a broader range of conditions and to further investigate the long-term effects of AMF inoculations on soil biodiversity and ecosystem functioning.