Effect of fire and post-fire management on soil microbial communities in a lower subtropical forest ecosystem after a mountain fire.

Wildfires and post-fire management exert profound effects on soil properties and microbial communities in forest ecosystems. Understanding microbial community recovery from fire and what the best post-fire management should be is very important but needs to be sufficiently studied. In light of these gaps in our understanding, this study aimed to assess the short-term effects of wildfire and post-fire management on both bacteria and fungi community composition, diversity, structure, and co-occurrence networks, and to identify the principal determinants of soil processes influencing the restoration of these communities. Specifically, we investigated soil bacterial and fungal community composition, diversity, structure, and co-occurrence networks in lower subtropical forests during a short-term (<3 years) post-fire recovery period at four main sites in Guangdong Province, southern China. Our results revealed significant effects of wildfires on fungal community composition, diversity, and co-occurrence patterns. Network analysis indicated reduced bacterial network complexity and connectivity post-fire, while the same features were enhanced in fungal networks. However, post-fire management effects on microbial communities were negligible. Bacterial diversity correlated positively with soil microbial biomass nitrogen, soil organic carbon, and soil total nitrogen. Conversely, based on the best random forest model, fungal community dynamics were negatively linked to nitrate-nitrogen and soil water content. Spearman's correlation analysis suggested positive associations between bacterial networks and soil factors, whereas fungal networks exhibited predominantly negative associations. This study elucidates the pivotal role of post-fire management in shaping ecological outcomes. Additionally, it accentuates the discernible distinctions between bacterial and fungal responses to fire throughout a short-term recovery period. These findings contribute novel insights that bear significance in evaluating the efficacy of environmental management strategies.

The rapid start-up of CANON process through adding partial nitration sludge to ANAMMOX system.

This study aimed to start up the completely autotrophic nitrogen removal over nitrite (CANON) process after adding partial nitration (PN) sludge to the ANAMMOX reactor, so as to help the rapid start-up and stable operation of the CANON process in practical engineering applications. There were three steps in the research: cultivating the PN sludge, building a reliable ANAMMMOX system, and finally starting and running the CANON process. The PN sludge was successfully cultivated in less than 45 days with around 90% nitrite accumulation rate. The ANAMMOX reactor enriched a significant quantity of red granular sludge within 70 days, achieving the maximum nitrogen removal rate of 1.74 kg/(m3·d). Eventually, the CANON reactor was started up successfully, which achieved 95.08% of average ammonium removal efficiency and 84.51% of average total nitrogen removal efficiency in 60 days. The residual recalcitrant nitrite-oxidizing bacteria in the CANON process was successfully inhibited by intermittent aeration and 12 mg/L free ammonia in UASB reactor. Besides, Candidatus Kuenenia, Candidatus Brocadia and Nitrosomonas were the main functional microorganisms involved in the CANON process.

The nitrogen removal performance and microbial community on mixotrophic denitrification process.

Sulfur autotrophic denitrifiers and heterotrophic denitrifiers widely exist in aquatic ecosystem, however, the response of sulfide to the microbial community structure in mixotrophic denitrification ecosystem is unknown yet. In this study, the denitrification performance and microbial community were explored by changing the molar ratio of influent C/N/S. From the level of genus, the joint action of Thauera, Pacacoccus, Fusibacter Pseudoxanthomonas, Thiobacillus, Sulfurovum and Sulfurimonas brought about the efficient denitrification performance in the mixotrophic system. Thauera increased from from 0.97% to more than 13%, and the relative abundances of Thiobacillus and Sulfurimonas were about 4.14% and 3.89% separately after adding S2-. The results of this study showed that the denitrification performance could be indeed intensified in the mixotrophic system, among which provided a theoretical basis for establishing an efficient biological nitrogen removal system.

A Multi-Scale and Multi-Level Fusion Approach for Deep Learning-Based Liver Lesion Diagnosis in Magnetic Resonance Images with Visual Explanation.

Many computer-aided diagnosis methods, especially ones with deep learning strategies, of liver cancers based on medical images have been proposed. However, most of such methods analyze the images under only one scale, and the deep learning models are always unexplainable. In this paper, we propose a deep learning-based multi-scale and multi-level fusing approach of CNNs for liver lesion diagnosis on magnetic resonance images, termed as MMF-CNN. We introduce a multi-scale representation strategy to encode both the local and semi-local complementary information of the images. To take advantage of the complementary information of multi-scale representations, we propose a multi-level fusion method to combine the information of both the feature level and the decision level hierarchically and generate a robust diagnostic classifier based on deep learning. We further explore the explanation of the diagnosis decision of the deep neural network through visualizing the areas of interest of the network. A new scoring method is designed to evaluate whether the attention maps can highlight the relevant radiological features. The explanation and visualization make the decision-making process of the deep neural network transparent for the clinicians. We apply our proposed approach to various state-of-the-art deep learning architectures. The experimental results demonstrate the effectiveness of our approach.

An Interpretation Architecture for Deep Learning Models with the Application of COVID-19 Diagnosis.

The Coronavirus disease 2019 (COVID-19) has become one of the threats to the world. Computed tomography (CT) is an informative tool for the diagnosis of COVID-19 patients. Many deep learning approaches on CT images have been proposed and brought promising performance. However, due to the high complexity and non-transparency of deep models, the explanation of the diagnosis process is challenging, making it hard to evaluate whether such approaches are reliable. In this paper, we propose a visual interpretation architecture for the explanation of the deep learning models and apply the architecture in COVID-19 diagnosis. Our architecture designs a comprehensive interpretation about the deep model from different perspectives, including the training trends, diagnostic performance, learned features, feature extractors, the hidden layers, the support regions for diagnostic decision, and etc. With the interpretation architecture, researchers can make a comparison and explanation about the classification performance, gain insight into what the deep model learned from images, and obtain the supports for diagnostic decisions. Our deep model achieves the diagnostic result of 94.75%, 93.22%, 96.69%, 97.27%, and 91.88% in the criteria of accuracy, sensitivity, specificity, positive predictive value, and negative predictive value, which are 8.30%, 4.32%, 13.33%, 10.25%, and 6.19% higher than that of the compared traditional methods. The visualized features in 2-D and 3-D spaces provide the reasons for the superiority of our deep model. Our interpretation architecture would allow researchers to understand more about how and why deep models work, and can be used as interpretation solutions for any deep learning models based on convolutional neural network. It can also help deep learning methods to take a step forward in the clinical COVID-19 diagnosis field.