晚中新世托尔通期（11.61~7.25 Ma）比现代更加温暖和潮湿，却有着与工业革命前相近的大气二氧化碳分压，这种低大气二氧化碳分压下的暖室气候在整个新生代都很特殊，搞清其驱动机制有助于预测未来气候变化。对此有两种解释：基于指标记录的晚中新世气候—二氧化碳分压“解耦假说”和基于数值模拟的“协同作用假说”。指标重建的地质记录表明晚中新世的气候可能并不受二氧化碳分压影响，即气候与二氧化碳分压发生解耦。数值模拟表明晚中新世异于现代的植被分布和地形构造等，很可能促成晚中新世全球温度的升高。但数值模拟很难充分模拟出晚中新世的升温幅度和模式。对晚中新世开展准确且高分辨率的二氧化碳分压重建是未来指标重建工作的重点，而植被反馈、云反馈、水蒸汽反馈和土壤性质是影响晚中新世暖室气候的主要因素，未来的数值模拟工作应朝这些方向推进。

The period of the late Miocene Tortonian （11.61~7.25 Ma） was warmer and wetter than today， with atmospheric partial pressure of carbon dioxide （pCO₂） near the preindustrial level. Greenhouse climate under low pCO₂ was rare throughout the Cenozoic and understanding its mechanisms will help to better forecast the future climate. We summarized two hypotheses to elucidate this mechanism. One is the late Miocene climate-pCO₂ "decoupling hypothesis" based on geological records， and the other is "synergistic effects hypothesis" based on climate modeling. Geological records indicate that the late Miocene climate may not have been affected by pCO₂， that is， climate and pCO₂ were decoupled. Climate modeling results indicate that different vegetation and tectonic conditions in the late Miocene may have contributed to the global temperature increase. However， realistically， it is difficult to fully simulate the amplitude and pattern of the late Miocene warmth. Future work should focus on reconstructing pCO₂ records with high accuracy and resolution. Vegetation， clouds， water vapor feedback， and soil properties may be the dominant factors contributing to the late Miocene greenhouse climate， which should also be considered in future simulation work.