Peatlands help regulate climate by sequestering (net removal) carbon from the atmosphere and storing it in plants and soils. However, as mean annual air temperature (MAAT) increases, peat carbon stocks may decrease. We conducted an in-depth synthesis of current knowledge about ecosystem controls on peatland carbon storage and fluxes to constrain the most influential parameters in probabilistic modelling of peat carbon sinks, such as Bayesian belief networks. Evaluated parameters included climate, carbon flux and mass, land cover, landscape position (defined here as elevation), fire records, and current and future climate scenarios for a 74,300 km2 landscape in the Hudson Bay Lowlands, Canada. The Bayesian belief network was constructed with four tiers: (1) exposure, expressed as MAAT, and the state variables of elevation and land cover; (2) sensitivity, expressed as ecosystem conditions relevant to peat carbon mass and its quality for decomposition, peat wetness, and fire; (3) carbon dioxide and methane fluxes and peat combustion; and (4) vulnerability of peat carbon sink strength under warmer MAAT. Simulations were conducted using current (-3.0 to 0.0 ᵒC), moderately warmer (0.1 to 4.0 ᵒC), and severely warmer (4.1 to 9.0 ᵒC) climate scenarios. Results from the severely warmer climate scenario projected an overall drying of peat, with approximately 20% reduction in the strong sink categories of net ecosystem exchange and peat carbon sink strength for the severely and, to a lesser degree, the moderately warmer climate scenarios relative to current MAAT. In the warmest temperature simulation, probability of methane emission decreased slightly and the probability of the strong peat carbon sink strength was 27% lower due to peat combustion. Our Bayesian belief network can assist land planners in decision-making for peatland-dominated landscapes, such as identifying high carbon storage areas and those projected to be at greatest risk of carbon loss due to climate change. Such areas may be designated, for example, as protected or reduced management intensity. The Bayesian belief network presented here is built on an in-depth knowledge synthesis to construct conditional probability tables, so is expected to apply to other peatland-dense jurisdictions by changing only elevation, peatland types, and MAAT.

中文翻译：

---

加拿大哈德逊湾低地的泥炭碳对预计气候变暖的脆弱性：泥炭地主导景观中土地利用规划的决策支持工具

泥炭地通过从大气中隔离（净去除）碳并将其储存在植物和土壤中来帮助调节气候。然而，随着年平均气温 (MAAT) 的增加，泥炭碳储量可能会减少。我们对有关泥炭地碳储存和通量的生态系统控制的当前知识进行了深入综合，以限制泥炭碳汇概率建模中最有影响的参数，例如贝叶斯信念网络。评估参数包括加拿大哈德逊湾低地 74,300 平方公里景观的气候、碳通量和质量、土地覆盖、景观位置（此处定义为海拔）、火灾记录以及当前和未来的气候情景。贝叶斯信念网络由四层构建：（1）暴露，表示为 MAAT，以及高程和土地覆盖的状态变量；(2) 敏感性，表示为与泥炭碳质量及其分解质量、泥炭湿度和火灾相关的生态系统条件；(3) 二氧化碳和甲烷通量和泥炭燃烧；(4) 温暖 MAAT 下泥炭碳汇强度的脆弱性。模拟是使用当前（-3.0 至 0.0 ᵒC）、中度温暖（0.1 至 4.0 ᵒC）和严重温暖（4.1 至 9.0 ᵒC）气候情景进行的。严重变暖气候情景的结果预测泥炭总体干燥，对于相对于 2000 年的严重气候情景和适度变暖的气候情景，净生态系统交换和泥炭碳汇强度的强汇类别减少约 20%。当前 MAAT。在最热的温度模拟中，由于泥炭燃烧，甲烷排放的概率略有下降，强泥炭碳汇强度的概率降低了 27%。我们的贝叶斯信念网络可以帮助土地规划者对以泥炭地为主的景观进行决策，例如确定高碳储存区域和预计因气候变化而面临最大碳损失风险的区域。例如，这些区域可以被指定为受保护的或降低的管理强度。此处介绍的贝叶斯信念网络建立在深入的知识综合之上，以构建条件概率表，因此有望通过仅改变海拔、泥炭地类型和 MAAT 来应用于其他泥炭地密集的管辖区。我们的贝叶斯信念网络可以帮助土地规划者对以泥炭地为主的景观进行决策，例如确定高碳储存区域和预计因气候变化而面临最大碳损失风险的区域。例如，这些区域可以被指定为受保护的或降低的管理强度。此处介绍的贝叶斯信念网络建立在深入的知识综合之上，以构建条件概率表，因此有望通过仅改变海拔、泥炭地类型和 MAAT 来应用于其他泥炭地密集的管辖区。我们的贝叶斯信念网络可以帮助土地规划者对以泥炭地为主的景观进行决策，例如确定高碳储存区域和预计因气候变化而面临最大碳损失风险的区域。例如，这些区域可以被指定为受保护的或降低的管理强度。此处介绍的贝叶斯信念网络建立在深入的知识综合之上，以构建条件概率表，因此有望通过仅改变海拔、泥炭地类型和 MAAT 来应用于其他泥炭地密集的管辖区。