• •    

标准化主轴回归在植物性状研究中的应用

王昊羽, 宗璐, 陈思翀   

  1. 中国科学院武汉植物园植物多样性与特色经济作物全国重点实验室, 湖北 430074 中国
    中国科学院大学, 北京 100049 中国
    香港大学生物科学学院, 香港特别行政区 999077 中国
  • 收稿日期:2026-04-20 修回日期:2026-08-04
  • 基金资助:
    国家自然科学基金(32371612); 国家自然科学基金(32401668)

Applications of standardised major axis regression in plant trait research

WANG Hao-Yu, ZONG Lu, CHEN Si-Chong   

  1. State Key Laboratory of Plant Diversity and Specialty Crops, Wuhan Botanical Garden, Chinese Academy of Sciences 430074, China
    , University of Chinese Academy of Sciences 100049, China
    School of Biological Sciences, The University of Hong Kong 999077, China
  • Received:2026-04-20 Revised:2026-08-04
  • Supported by:
    Support by the National Science Foundation of China(32371612); Support by the National Science Foundation of China(32401668)

摘要: 标准化主轴(Standardised major axis, SMA)回归作为一种典型的模型II回归方法,能够同时处理两个变量的变异特征,已成为量化植物性状异速生长关系的关键统计工具。然而,当前对其基本原理与方法体系仍存在一定认识偏差。本文系统梳理了SMA回归的理论基础、分析流程及其在植物生态学中的应用进展。首先,阐述了SMA回归的理论框架,并结合R语言(如smatr包)介绍其实际操作流程;其次,本文进一步综述了SMA在植物性状研究中的应用,包括植物器官大小与个体生物量之间的异速生长关系、不同器官间资源分配及协同权衡关系以及器官内部功能性状(如叶片功能性状经济谱)的内在联系。最后,本文指出当前研究在对数转换后线性假设适用性、非线性主轴回归、多层级数据整合及繁殖器官研究等方面仍存在局限,并提出应加强多维数据整合与跨尺度验证。

关键词: 模型II回归, 功能性状, 生物标度, 异速生长, 资源分配, 线性回归

Abstract: Standardised major axis (SMA) regression, a typical Model II regression approach, explicitly accounts for variation in both variables and has become a key statistical tool for quantifying allometric relationships among plant traits. However, its underlying principles and methodological framework remain subject to some misunderstandings. Here, we provide a systematic synthesis of the theoretical basis, analytical workflow, and applications of SMA in plant ecology. First, we outline the theoretical framework of SMA regression and introduce its practical implementation in R (e.g. the smatr package). We then review its applications in plant trait research, including allometric relationships between organ size and whole-plant biomass, patterns of resource allocation and trade-offs among organs, and the internal associations of functional traits within organs (e.g. leaf economic spectrum). Finally, we highlight current limitations, including the applicability of log-transformed linearity assumptions, non-linear major axis regression, the integration of multi-level data, and the relative scarcity of studies on reproductive organs. We propose that future work should emphasise multidimensional data integration and cross-scale validation.

Key words: Model II regression, functional trait, biological scaling, allometry, resource allocation, linear regression