关键词: 常绿阔叶林, 环境因子, 净初级生产力, 降水利用效率, 随机森林

Abstract: Evergreen broadleaved forests (EBFs) play a crucial role in maintaining biodiversity and ecosystem functions, which are highly sensitive to climate change and human activities. However, global spatial distributions and the underlying mechanisms of net primary productivity (NPP) and precipitation use efficiency (PUE) in EBFs remain poorly understood. Here, we integrated measured NPP data from 272 EBFs sites worldwide with a random forest model to predict global NPP and PUE patterns and to evaluate climatic and edaphic drivers. The results showed that the predicted global NPP values ranged from 909.08 to 3213.11 g C·m^-2·a^-1, while PUE varied from 0.22 to 5.79 g C m^-2·mm^-1. NPP and PUE exhibited similar global distribution patterns, both increasing at forests below an elevation of 400 m and then declining above this threshold. Temperature and precipitation were the principal determinants of NPP and PUE variations, whereas soil physical properties played a relatively minor role. We identified the threshold effects in these relationships, including a shift in the NPPprecipitation relationship at 1200 mm, with precipitation exerting a stronger influence on NPP at temperatures below 19 ℃. PUE displayed threshold responses to temperature and precipitation, with higher values being observed at precipitations below 2500 mm or temperatures below 24 ℃. The combined effects of precipitation and temperature on both NPP and PUE depend on their interactive influence rather than the isolated effects. These findings provide critical parameters for carbon-water cycle models in EBFs and enhance the predictive capabilities under future climate change scenarios.

Key words: evergreen broadleaved forest, environmental factor, net primary productivity, precipitation use efficiency, random forest