Climate and Edaphic Factors Partly Explain Leaf Litterfall Rates of Species From Tropical Semi-Deciduous Ecosystems

Carbon cycling in semi-deciduous ecosystems is difficult to observe because plant species shed leaves for only a short period (4-6 weeks), and some remain dormant for only a few weeks. These short phenological phases limit ecosystem studies. To our knowledge, no database currently describes spatial and temporal patterns of litterfall in semi-deciduous ecosystems. Therefore, this study aimed to identify factors associated with leaf litterfall rates in these ecosystems. We compiled leaf litterfall data from 31 global semi-deciduous sites in the literature and collected climatic data (precipitation, air temperature, vapor pressure deficit, radiation, daylength) and edaphic data (available water storage capacity, top-soil bulk density, clay cation exchange capacity, clay and sand fractions, pH, and soil carbon). Models of linear mixed effects (LMEM) and random forest were used to examine climatic and soil controls on litterfall rates. Radiation, daylength, rainfall, and soil pH explained 51% of the variation in litterfall rates across the data set that was used for calibration. All the predictors were negatively correlated with litterfall. The LMEM and random forest model revealed consistent key predictors of the litterfall. When the developed model was applied at independent sites, the explanatory power of the model was 24%, suggesting the need to include biotic factors such as species composition, vegetation structure, and functional traits in the model. Our synthesized datasets can be used to calibrate litterfall rates that are predicted using process-based models, such as land surface models.