Bacterial biomarkers are linked more closely to wheat yield formation than overall bacteria in fertilized soil

Abstract Soil bacteria play pivotal roles in agroecosystem functioning. However, the immense diversity of soil bacterial communities masks the effects of some species, such as bacterial biomarkers, on soil nutrient cycling and crop growth. We analyzed biological and chemical soil parameters of five fertilization treatments (CK, LN, CN, HN, LNM: 0, 150, 200, 250 kg N hm −2 , 150 kg N hm −2 + 15,000 kg hm −2 sheep manure). The random forest models were used to select bacterial biomarkers and assess the contributions of bacterial biomarkers and overall bacteria to wheat yield. Moreover, we used a partial least squares path modelling to explore the relationship between fertilization, bacterial biomarkers, carbon (C) and nitrogen (N) cycling functions, soil nutrients, and wheat yield. The results showed that bacterial biomarkers are better predictors of wheat yield than overall bacteria, explaining 52.54% and 16.00% of the variation in yield, respectively. Bacterial biomarkers affected wheat production by increasing soil organic carbon, and available nitrogen content. The LNM treatment significantly improved the relative abundance of beneficial biomarkers, such as Sphingomonadales, Xanthomonadaceae, Lysobacter, and Streptomyces, compared to the LN treatment. In addition, the LNM treatment promoted chitinlysis, cellulolysis, xylanolysis, aromatic hydrocarbon degradation, and aerobic ammonia oxidation, relative to the other inorganic N treatments (LN, CN, and HN). Our results emphasize the role of bacterial biomarkers in wheat yield formation, providing new insights into maintaining agricultural productivity by taking advantage of soil bacterial biomarkers.