Excessive nutrient and sediment inputs threaten ecological condition in many estuaries. We describe a Bayesian Belief Network (BBN) that calculates an Estuary Trophic Index (ETI) score ranging between 0 (no symptoms of eutrophication) to 1 (grossly eutrophic) for estuaries in Aotearoa New Zealand (NZ). The ETI BBN includes estuary physiographic characteristics (estuary type, flushing time, intertidal area, estuary closure state, water column stratification) and nutrient and sediment loads available from existing geospatial tools and databases, that drive responses of ‘primary’ indicators (macroalgae and phytoplankton biomass) and ‘secondary’ indicators (or symptoms) of estuary ecological impairment (sediment carbon, sediment apparent redox potential discontinuity depth, water column oxygen, macrobenthos and seagrass condition). Relationships between the BBN nodes are based primarily on observational and model-based information from NZ and international studies rather than expert opinion. The model can be used in a purely predictive manner under knowledge-poor situations, using only the physiographic drivers and nutrient/sediment loads, or refined using field-derived observations of indicator values to reduce the uncertainty associated with the probabilistic BBN score. It is designed for shallow tidal lagoons, tidal river estuaries and coastal lakes; systems which are sensitive to eutrophication and sedimentation pressure and are common in NZ and globally. Modelled ETI BBN scores agreed well with ETI scores calculated from observed indicator values for 11 well-studied NZ estuaries. We predict ecological condition of 291 NZ estuaries, most of which have no monitored information on trophic state. We illustrate capabilities of the ETI BBN with two case studies: to evaluate improvements in estuary health arising from diversion of wastewater from an estuary via an ocean outfall, and to estimate catchment diffuse nutrient load reductions required to meet estuary health objectives. The ETI BBN may serve as a template for other agencies wishing to develop similar tools.

过多的养分和沉积物输入威胁着许多河口的生态条件。我们描述了一个贝叶斯信念网络 (BBN)，它计算了新西兰 (NZ) 河口的河口营养指数 (ETI) 分数，范围在 0（没有富营养化症状）到 1（严重富营养化）之间。ETI BBN 包括河口自然地理特征（河口类型、冲刷时间、潮间带区域、河口封闭状态、水柱分层）以及现有地理空间工具和数据库中可用的养分和沉积物负荷，它们推动了“主要”指标（大型藻类和浮游植物）的响应生物量）和河口生态损害的“次要”指标（或症状）（沉积物碳、沉积物表观氧化还原电位间断深度、水柱氧、大型底栖生物和海草状况）。BBN 节点之间的关系主要基于来自新西兰和国际研究的观察和基于模型的信息，而不是专家意见。该模型可以在知识匮乏的情况下以纯粹的预测方式使用，仅使用自然地理驱动因素和养分/沉积物负荷，或使用现场派生的指标值观察结果进行改进，以减少与概率 BBN 分数相关的不确定性。专为浅潮汐泻湖、潮汐河流入海口和沿海湖泊而设计；对富营养化和沉降压力敏感的系统在新西兰和全球都很常见。建模的 ETI BBN 分数与根据 11 个经过充分研究的新西兰河口的观察指标值计算的 ETI 分数非常吻合。我们预测 291 个新西兰河口的生态状况，其中大多数没有关于营养状态的监测信息。我们通过两个案例研究来说明 ETI BBN 的能力：评估因河口分流废水而对河口健康的改善通过海洋排放口，并估计满足河口健康目标所需的集水区扩散营养负荷减少。ETI BBN 可以作为希望开发类似工具的其他机构的模板。