Montane cloud forests (MCFs) are rare and fragile ecosystems. In Taiwan, these forests are distributed across mountainous areas throughout the island. Situated at the junction of tropical and subtropical zones, as well as tectonic plate boundaries, Taiwan features diverse climatic zones and numerous mountain ranges. The vegetation is further influenced by occasional typhoons and regular monsoons, and also by strong fog frequency. This study aims to compare the differences among MCFs in different regions of Taiwan by examining species diversity and community assembly processes.

Using 1751 cloud forest vegetation plots from the National Vegetation Dataset of Taiwan, we categorized the island into four regions: north (491 plots), south (397 plots), east (376 plots), and west (487 plots) based on climatic classification. Different alpha diversity indices for each region were calculated by rarefaction to 752 plots (using the iNEXT method). To quantify the relative proportion of deterministic and stochastic assembly processes in each region, we repeatedly subset 100 plots in each region and analyzed variance partitioning between environmental variables representing a deterministic part, and spatial variables (Moran’s Eigenvector Maps derived from geographic coordinates) representing a stochastic part. After variance partitioning, we used the rdacca.hp package to determine the relative importance of each deterministic factor.

The results indicate that MCFs in southern Taiwan exhibit the highest species diversity, while western Taiwan shows the lowest. The influence of deterministic processes in western Taiwan is significantly higher than in other regions, whereas stochastic processes show no significant difference across the four regions. Further analysis of deterministic factors revealed that mean annual temperature is the most critical factor across all regions. Strong environmental selection, primarily driven by mean annual temperature, limits the vegetation to species capable of adapting to these specific conditions. Stochastic processes explain a portion of the vegetation composition related to spatial positioning, but their impact does not vary significantly across regions. This may be because MCFs in Taiwan are situated within a narrow elevational belt despite their broad geographical distribution.

Although this study is based on the analysis of existing datasets and their relationship with environmental and spatial factors, we speculate that similar patterns of community assembly would likely be observed in studies focusing on temperature gradients and fog frequency in other comparable island ecosystems. This study provides a new ecological interpretation of the roles of temperature and fog gradients, offering valuable insights for the future conservation of island montane cloud forests.