Early season soil microbiome best predicts wheat grain quality

Abstract

Previous studies have shown that it is possible to accurately predict wheat grain quality and yields using microbial indicators. However, it is uncertain what the best timing for sampling is. For optimal usefulness of this modeling approach, microbial indicators from samples taken early in the season should have the best predictive power. Here, we sampled a field every two weeks across a single growing season and measured a wide array of microbial parameters (amplicon sequencing, abundance of N-cycle related functional genes, and microbial carbon usage) to find the moment when the microbial predictive power for wheat grain baking quality is highest. We found that the highest predictive power for wheat grain quality was for microbial data derived from samples taken early in the season (May–June) which coincides roughly with the seedling and tillering growth stages, that are important for wheat N nutrition. Our models based on LASSO regression also highlighted a set of microbial parameters highly coherent with our previous surveys, including alpha- and beta-diversity indices and N-cycle genes. Taken together, our results suggest that measuring microbial parameters early in the wheat growing season could help farmers better predict wheat grain quality.

Long-Term Tillage and Crop Rotation Regimes Reshape Soil-Borne Oomycete Communities in Soybean, Corn, and Wheat Production Systems

“…In this study, sampling was conducted in the fall, when a majority of the available plant tissue in the soil was dead, favoring saprophytic oomycetes such as Pythium and Globisporangium . Asad et al [ 66 ] found that microbiome sampling early in the growing season was closely related to final seed quality. Therefore, it is likely that sampling oomycete communities in the spring or summer may result in different trends and provide further insights into their dynamics throughout the growing season.…”

The forecasting power of the microbiome

“…Similarly, shrinkage or regularisation methods, such as ridge, LASSO, or elastic net, include all predictors in the equation but reduce the size of the coefficient estimates towards zero. It was recently used to compare the accuracy of forecasting wheat yield and quality from microbiome datasets collected at different dates [36]. However, for many of the sampling dates close to harvest, the LASSO procedure resulted in a null model, where all coefficients were shrunk to zero, suggesting that the forecasting potential of the microbiome at these dates was mediocre, or that LASSO was not the most appropriate tool for this task [36].…”

“…It was recently used to compare the accuracy of forecasting wheat yield and quality from microbiome datasets collected at different dates [36]. However, for many of the sampling dates close to harvest, the LASSO procedure resulted in a null model, where all coefficients were shrunk to zero, suggesting that the forecasting potential of the microbiome at these dates was mediocre, or that LASSO was not the most appropriate tool for this task [36]. Similarly, LASSO and elastic net resulted in less accurate disease prediction models than when using random forest selection, even though the accuracy of the latter approach was optimal when more than 60 species were selected [21], which is still a lot of descriptors.…”

Soil fungal community structure and function response to rhizoma perennial peanut cultivars

“…Crop-associated microbiomes play an important role in regulating many key ecological processes, including carbon (C) and nitrogen (N) cycling [ 2 ], nutrient acquisition [ 3 ], and soil formation [ 4 ], consequently affecting crop growth and health [ 5 ]. In a recent study, early season soil microbiome has been shown a better ability for predicting wheat grain quality [ 6 ]. Soil fungi, such as Penicillium [ 7 ], Trichoderma [ 8 ], and arbuscular mycorrhiza fungus (AMF) [ 9 ], can promote plant growth.…”