Development of models for estimating leaf chlorophyll and nitrogen contents in tree species with respect to seasonal changes

Models were developed to estimate nondestructively chlorophyll (Chl) content per unit of leaf area (Chlₐᵣₑₐ) and nitrogen content per unit of leaf area (Nₐᵣₑₐ) using readings of two optical meters for five warm-temperate, evergreen, broadleaved tree species (Castanopsis sieboldii, Cinnamomum tenuifolium, Eurya japonica, Machilus thunbergii, and Neolitsea sericea). It was determined whether models should be adjusted seasonally. Readings (were obtained six times during a year period and Chlₐᵣₑₐ and Nₐᵣₑₐ were determined using destructive methods. Bayesian inference was used to estimate parameters of models that related optical meter readings to Chlₐᵣₑₐ or Nₐᵣₑₐ for each species. Deviance information criterion values were used to select the best among models, including the models with seasonal adjustment. The selected models were species-specific and predicted Chlₐᵣₑₐ accurately (R ² = 0.93–0.96). The best model included parameters with seasonal adjustments for one out of five species. Model-based estimates of Nₐᵣₑₐ were not as accurate as those for Chlₐᵣₑₐ, but they were still adequate (R ² = 0.64–0.82). For all species studied, the best models did not include parameters with seasonal adjustments. The estimation methods used in this study were rapid and nondestructive; thus, they could be used to assess a function of many leaves and/or repeatedly on individual leaves in the field.