ESTIMATION OF LEAF AREA, LEAF MASS AND SPECIFIC LEAF AREA FOR TREES OF DIFFERENT LIFE-FORMS IN A KARST FOREST BASED ON LINEAR MIXED-EFFECTS MODELS

作者：Wu, Q. C.; Luo, G. L.; Qi, T.; Qi, Y. J.^*

来源：APPLIED ECOLOGY AND ENVIRONMENTAL RESEARCH, 2022, 20(3): 2017-2033.

DOI：10.15666/aeer/2003_20172033

• leaf structure parameters
• seasonal variations
• horizontal directions of canopy
• random effects
• non-destructive measurement

Leaves act as an important bridge between plants and the external environment. Many studies have been conducted to establish more accurate and efficient models for predicting values of leaf traits. However, the model based on tree species is not representative enough for tropical and subtropical forests with abundant tree species and complex structures. Additionally, ordinary models are generally insufficient to describe the spatial and temporal changes of leaves because of the variations between trees. Based on linear mixed-effects models (LMM), we estimated the leaf area (LA), leaf mass (LM), and specific leaf area (SLA) for tree species of four life forms in a karst primary forest. Our results suggested that LMM were reasonable and accurate in fitting and predicting LA and LM for different life forms in different seasons. The most accurate predictions were obtained while using the product of leaf length and leaf width. Specifically, LMM performed better (R-2 = 0.92 to 0.99 and AIC = 118.5 to 4306.76 for leaf area; R-2 = 0.88 to 0.94 and AIC = -4389.5 to -969.1 for leaf mass) than the models only considered the fixed effects (R-2 = 0.89 to 0.99 and AIC = 119.1 to 4464.79 for leaf area; R-2 = 0.79 to 0.87 and AIC = -3179.7 to -940.2 for leaf mass). The mean absolute error percent values were 0.9%-14.4% for leaf area and 1.1% to 17.1% for leaf mass for four life forms. Considering the accuracy of the models and the sampling effort, the optimal number of sample leaves for SLA estimation was about 60-80.

• 单位
  贵州大学