作者：Guo, J., Zhao, J., Yu, C., Qi, W., Qu, J., Duan, Y. & Liu, H.

Algal volatile organic compounds (AVOCs) act as real-time metabolic signals that enable accurate bloom prediction in single-species systems. However, interspecies interactions reshape algal growth dynamics across bloom stages, and evaluating how AVOC-based warning approaches perform under algal interactions is crucial. Here, proton transfer reaction time-of-flight mass spectrometry (PTR-TOF-MS) was employed to profile AVOCs from monocultures and cocultures of Microcystis aeruginosa and Chlorella vulgaris, and machine learning models were established to predict algal density. Coculture conditions, which inhibited C. vulgaris but stimulated M. aeruginosa, induced AVOCs putatively associated with fatty acid beta-oxidation and carotenoid degradation, aligning with reported oxidative stress and intercellular signaling responses. These AVOCs (n = 1268) exhibited remarkable predictive power for algal density under interspecies interactions, with an extreme gradient boosting model achieving high accuracy (R 2: 0.96). Shapley additive explanation analysis identified two categories of predictive AVOCs: fundamental metabolism-associated (e.g., methanethiol and dimethylamine, 22.66%) and interactions-associated (e.g., isophorone and (3E)-4,8-dimethyl-1,3,7-nonatriene (DMNT), 6.97%) AVOCs. Interaction-associated AVOCs, particularly terpenoids known to participate in stress signaling and interspecific communication, exhibited a 123% increase in predictive importance. Transcriptomic analyses suggested that fundamental metabolism-associated AVOCs were functionally connected to photosynthesis and ribosomal pathways, whereas interaction-associated AVOCs were more closely related to carotenoid and fatty acid metabolic processes. The significant correlation between DMNT content and algal biomass was validated in a natural lake. These findings provide a biologically grounded framework for real-time bloom prediction by incorporating interaction-mediated volatiles into early warning strategies.

藻类挥发性有机物（AVOCs）可作为实时代谢信号，在单物种体系中实现准确的藻华预测。然而，种间相互作用会在藻华不同阶段重塑藻类生长动态，因此评估基于AVOC的预警方法在藻类相互作用下的表现至关重要。研究采用质子转移反应飞行时间质谱（PTR-TOF-MS）解析了铜绿微囊藻（Microcystis aeruginosa）与普通小球藻（Chlorella vulgaris）在单培养与共培养体系中的AVOCs谱，并建立了预测藻类密度的机器学习模型。共培养条件抑制了小球藻但促进了铜绿微囊藻的生长，并诱导产生推定与脂肪酸β-氧化及类胡萝卜素降解相关的AVOCs，这与已报道的氧化应激及胞间信号响应相吻合。这些AVOCs（n = 1268）在种间相互作用条件下对藻类密度展现出显著的预测能力：极端梯度提升模型达到高精度（R ² = 0.96）。SHAP（Shapley加性解释）分析识别出两类具有预测价值的AVOCs：基础代谢相关类（如甲硫醇、二甲胺，重要性占比22.66%）与相互作用相关类（如异佛尔酮、(3E)-4,8-二甲基-1,3,7-壬三烯（DMNT），占比6.97%）。其中相互作用相关AVOCs——尤其是参与胁迫信号传导与种间通讯的萜类物质——的预测重要性增幅达123%。转录组分析表明，基础代谢相关AVOCs在功能上与光合作用及核糖体通路相关联，而相互作用相关AVOCs则与类胡萝卜素及脂肪酸代谢过程关系更为密切。天然湖泊中DMNT含量与藻类生物量的显著相关性进一步得到验证。研究通过将相互作用介导的挥发性信号纳入早期预警策略，为藻华的实时预测提供了一个有生物学依据的框架。

（来源：Environmental Science & Technology 2026 Issue 17  DOI: 10.1021/acs.est.5c18716）