Scholarlink

<p>Microbial communities are central to the functioning and resilience of biological water treatment systems, yet their structural and functional determinants remain poorly understood. Here, an interpretable machine-learned framework is developed to decode ecological drivers of treatment performance across 648 globally distributed biofilters. Keystone taxa, such as <i>Nitrospira</i>, <i>Hyphomicrobium</i>, <i>Flavobacterium</i>, are identified via deep ecological modeling. Models that include microbial ecological indicators, particularly the presence ratios of structural and functional keystone taxa, improve pollutant removal prediction (R² = 0.742) by 36.9% compared with models based only on process parameters. Generalized additive models with structured interactions further reveal volume, temperature, media size, and hydraulic retention time as ecological levers shaping microbial structure and function. Lab-scale biofilter experiments demonstrate consistent microbial shifts and improved treatment performance aligned with model predictions. These findings enable targeted microbiome control in biofilters and provide a foundation for adaptive operation of biological water systems under environmental variability.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Ecological levers for microbially driven water treatment enhance pollutant removal prediction

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月11日

查看全文
  • Lili Jin,
  • Jiayi Zhang,
  • Han Zhao,
  • Rui Ma,
  • Hui Huang,
  • Hongqiang Ren

摘要

Microbial communities are central to the functioning and resilience of biological water treatment systems, yet their structural and functional determinants remain poorly understood. Here, an interpretable machine-learned framework is developed to decode ecological drivers of treatment performance across 648 globally distributed biofilters. Keystone taxa, such as Nitrospira, Hyphomicrobium, Flavobacterium, are identified via deep ecological modeling. Models that include microbial ecological indicators, particularly the presence ratios of structural and functional keystone taxa, improve pollutant removal prediction (R² = 0.742) by 36.9% compared with models based only on process parameters. Generalized additive models with structured interactions further reveal volume, temperature, media size, and hydraulic retention time as ecological levers shaping microbial structure and function. Lab-scale biofilter experiments demonstrate consistent microbial shifts and improved treatment performance aligned with model predictions. These findings enable targeted microbiome control in biofilters and provide a foundation for adaptive operation of biological water systems under environmental variability.