Traditional population viability analysis (PVA) does not address the degree of measurement error or spatial and temporal variability of vital rate parameters, potentially leading to inappropriate conservation decision-making. We provide a methodology of applying Bayesian network (BN) modeling to PVA addressing these considerations, particularly for species with complex stage-class structures. We provide examples of three species from eastern Australia - hip pocket frog (Assa darilingtoni), squirrel glider (Petaurus norfolcensis) and giant burrowing frog (Heleioporus australiacus), comparing traditional matrix-based PVA with BN model analyses of mean stage abundance, quasi-extinction probability, and interval threshold extinction risk. Both approaches project similar population sizes, but BN PVA gave more clearly identifiable thresholds of population changes and extinction levels. The PVA BN uniquely represents complex stage-class structures and in a single network, including variation and uncertainty propagation of vital rates, to better inform conservation management decisions.

传统的种群生存力分析 (PVA) 没有解决测量误差的程度或生命率参数的空间和时间可变性，可能会导致不适当的保护决策。我们提供了一种将贝叶斯网络 (BN) 建模应用于 PVA 的方法，以解决这些考虑因素，特别是对于具有复杂阶段级结构的物种。我们提供了来自澳大利亚东部的三种物种的例子——臀袋蛙 ( Assa darilingtoni ) 、松鼠滑翔机 ( Peaurus norfolcensis)和巨型穴居蛙 ( Heleioporus australiacus)，比较传统的基于矩阵的 PVA 与 BN 模型分析的平均阶段丰度、准灭绝概率和间隔阈值灭绝风险。两种方法预测的种群规模相似，但 BN PVA 给出了更清晰可识别的种群变化和灭绝水平阈值。PVA BN 独特地表示复杂的阶段级结构和单个网络，包括生命率的变化和不确定性传播，以更好地为保护管理决策提供信息。