Canopy phenology is often modeled as the result of strong environmental limitation to maintaining leaves. However, diverse patterns of canopy phenology have been observed in tropical forests, even in tropical moist forests with weak environmental seasonality. More interestingly, coexisting plants of different species and sizes can show distinctive phenological seasonality in leaf flushing and shedding within the same site. However, in terrestrial biosphere/vegetation models, canopy phenology is often attributed to each plant functional type based on functional traits. This simplification can greatly underestimate the seasonality of vegetation/carbon dynamics in tropical forests and bias model predictions under changed seasonal climate and forest composition.
An alternative approach is to treat phenology as a strategy to optimize plant carbon gain, in which leaf functional traits, and leaf ageing processes play an important role in addition to environmental stress. Here, we tested the hypothesis that tropical canopy phenology is a result of maximizing leaf-level carbon gain under a seasonal climate. We developed a simple trait-driven numerical model to simulate canopy phenology based on carbon optimization. We first evaluated the model at the species level with 44 coexisting species at Barro Colorado Island (BCI), for which functional traits and long term weekly leaf litter fall data are available. We then evaluated the model at the ecosystem level with ecosystem-level photosynthetic seasonality and leaf litter seasonality across different Amazonian sites. Finally, we conducted numerical experiments by eliminating seasonality in temperature, vapor pressure deficit, and radiation to investigate which environmental factors are the main drivers of canopy phenological seasonality in the model.
Results/Conclusions
Preliminary results show that the carbon optimization framework for phenology can largely reproduce observed seasonal leaf litter fall for species at BCI for both evergreen and deciduous species. In particular, by considering leaf ageing, the model simulated that evergreen species can avoid environmental stress by exchanging leaves in the middle of the dry season. There are significant model-data mismatches for several species , whose canopy phenology might be influenced by reproduction phenology and plant mortality. At the ecosystem scale, the simulated monthly patterns of canopy photosynthetic capacity and leaf litterfall match well with the observations at two Amazonian sites. Numerical experiments suggest that seasonality in atmospheric water demand instead of radiation is likely the dominant driver in the optimization framework. Collectively, our results reveal that phenological patterns are likely to emerge from carbon optimization constrained by both leaf structural traits and the abiotic environment.