Principal components analysis of leaf economics traits for Sepilok (16 species).

Corresponding analyses for La Chonta (35 species; 18, 38) and San Lorenzo (14 species; 17). All analyses were performed on log-10 transformed variables across species. Two significant components explained a cumulative variation of 90.3% in Sepilok. Variation explained by individual components is provided for each trait. In the corresponding analysis for La Chonta, the second axis was marginally significant (standard deviation of 0.913); combined these two axes explained 82.5% of trait variation. Corresponding analysis using five of six traits (Rmass not available) from San Lorenzo; PCA axis 2 was not significant. In all three tables, leaf lifespan and LMA positively correlate with axis 1 while remaining traits are negatively correlated with axis 1. Foliar P concentration loads very strongly onto the second axis compared with all other significant traits in the Malaysian and Bolivian forests but less so in the Panamanian forest.

本研究针对La Chonta（35个物种；参考文献18、38）与San Lorenzo（14个物种；参考文献17）开展了对应分析。所有分析均基于各物种经以10为底的对数转换后的变量完成。在西必洛（Sepilok）样地的分析中，两个显著主成分累计解释了90.3%的性状变异，各主成分对单个性状的解释变异量已逐一列出。在La Chonta样地的对应分析中，第二轴呈边际显著（标准差为0.913）；两轴累计解释了82.5%的性状变异。针对San Lorenzo样地的对应分析仅采用了6个性状中的5个（Rmass数据不可得），其主成分分析（Principal Component Analysis, PCA）第二轴不显著。在全部三张表格中，叶片寿命与叶面积干质量（Leaf Mass per Area, LMA）均与第一轴呈正相关，其余性状均与第一轴呈负相关。相较于马来西亚与玻利维亚森林中的其余所有显著性状，叶片磷浓度在第二轴上的载荷极高，但在巴拿马森林中该效应相对较弱。