The Dynamic Changes of Brassica napus Seed Microbiota across the Entire Seed Life in the Field.

The seed microbiota is an important component given by nature to plants, protecting seeds from damage by other organisms and abiotic stress. However, little is known about the dynamic changes and potential functions of the seed microbiota during seed development. In this study, we investigated the composition and potential functions of the seed microbiota of rapeseed (Brassica napus). A total of 2496 amplicon sequence variants (ASVs) belonging to 504 genera in 25 phyla were identified, and the seed microbiota of all sampling stages were divided into three groups. The microbiota of flower buds, young pods, and seeds at 20 days after flowering (daf) formed the first group; that of seeds at 30 daf, 40 daf and 50 daf formed the second group; that of mature seeds and parental seeds were clustered into the third group. The functions of seed microbiota were identified by using PICRUSt2, and it was found that the substance metabolism of seed microbiota was correlated with those of the seeds. Finally, sixty-one core ASVs, including several potential human pathogens, were identified, and a member of the seed core microbiota, Sphingomonas endophytica, was isolated from seeds and found to promote seedling growth and enhance resistance against Sclerotinia sclerotiorum, a major pathogen in rapeseed. Our findings provide a novel perspective for understanding the composition and functions of microbiota during seed development and may enhance the efficiency of mining beneficial seed microbes.

Advanced nutrient removal and external sludge reduction: Demonstration in a pilot-scale sequencing batch reactor.

Although the addition of excess sludge fermentation products to improve nutrient removal from sewage is cost-effective, its application has rarely been demonstrated. In this study, the external sludge was first collected and fermented under a sludge retention time of 10 days, then introduced into SBR with a 1:15 sewage ratio. The results revealed a gradual increase in the nitrite accumulation ratio to 34.7% in the SBR at the end of the oxic stage after 64 days of adding fermented sludge products. In addition, the average effluent total nitrogen and phosphorous decreased to 7.3 and 0.5 mg/L, corresponding to removal efficiencies of 86.7% and 95.5%, respectively. On the other hand, the use of the fermented sludge products as external organic carbon sources in the SBR increased the external sludge reduction ratio to 42.5%. High-throughput sequencing demonstrated that the increase in the endogenous denitrifier community, polyphosphate-accumulating organisms, and fermentation bacteria were the main factors contributing to the increase in nutrient removal and excess sludge reduction. The economic evaluation indicated that the operational cost of the pilot-scale system saves 0.011$/m3 of sewage treated. PRACTITIONER POINTS: Fermented sludge addition effectively enhanced nutrient removal in pilot-scale SBR. Average effluent TN and PO4 3- -P decreased to 7.3 and 0.5 mg/L, respectively. Highest external sludge reduction rate was 42.5% in pilot-scale reactor. Sewage treatment cost can save 0.011$/m3 under advanced nutrient removal.

Ensemble Neural Networks (ENN): A gradient-free stochastic method.

In this study, an efficient stochastic gradient-free method, the ensemble neural networks (ENN), is developed. In the ENN, the optimization process relies on covariance matrices rather than derivatives. The covariance matrices are calculated by the ensemble randomized maximum likelihood algorithm (EnRML), which is an inverse modeling method. The ENN is able to simultaneously provide estimations and perform uncertainty quantification since it is built under the Bayesian framework. The ENN is also robust to small training data size because the ensemble of stochastic realizations essentially enlarges the training dataset. This constitutes a desirable characteristic, especially for real-world engineering applications. In addition, the ENN does not require the calculation of gradients, which enables the use of complicated neuron models and loss functions in neural networks. We experimentally demonstrate benefits of the proposed model, in particular showing that the ENN performs much better than the traditional Bayesian neural networks (BNN). The EnRML in ENN is a substitution of gradient-based optimization algorithms, which means that it can be directly combined with the feed-forward process in other existing (deep) neural networks, such as convolutional neural networks (CNN) and recurrent neural networks (RNN), broadening future applications of the ENN.