全球变化知识资源中心

The high diversity of plant species in the tropics has revealed complex phenological patterns and reproductive strategies occurring throughout the year. Describing and analysing tropical plant phenology, and detecting triggers, demands to consider the circular nature of recurrent life cycle events and the use of appropriated statistical metrics. Here, we explore analytical pitfalls potentially affecting results of studies that do not consider the circular nature of phenology data when comparing resting and non-resting systems, especially when accounting for phylogeny. We discuss definitions of the widely used first flowering date and revisit the literature on phylogenetic signal in plant phenology. We compare statistical analyses for tropical and temperate phenology by simulating communities with known phenological and phylogenetic structures. We demonstrate that ignoring the circular nature of phenological data underestimates the phylogenetic signal in plant phenology. Using the proposed circular transformation for non-resting tropical ecosystems and resting temperate systems prevented errors, yielding precise comparisons. Synthesis. The analysis of both non-resting and resting systems must consider the circularity of phenological events. Circular statistics is the appropriate approach to calculate phenological parameters, identify phylogenetic signal and assess drivers, allowing accurate cross-comparisons of phenology across environments at large spatial scales.