Data from: Lost in the other half: improving accuracy in geometric morphometric analyses of one side of bilaterally symmetric structures

Systematists and evolutionary biologists have widely adopted Procrustes-based geometric morphometrics for measuring size and shape in biology. Many structures, and in fact most animals, are bilaterally symmetric with an internal plane of symmetry (also called object symmetry). Often, when quantifying asymmetric variation is not an aim, only one or the other side is measured and analysed. This approach has been used in hundreds of studies. Its implicit assumption is that the information on the other side is redundant and a single side will therefore produce results mirroring those one would have obtained from the analysis of the entire structure with all its left and right landmarks. However, the extent to which this assumption is met has, to my knowledge, never been explored. Using two example datasets, I will show that congruence may be high in analyses at a macro-evolutionary level but much lower at a micro-evolutionary one, and inaccuracies might especially affect shape. I will discuss some of the other factors that may influence results and will suggest a simple expedient that can improve both the visualization and accuracy of shape analyses in one-side-only studies.

系统分类学家与进化生物学家已广泛采用基于普氏叠加的几何形态测量学（Procrustes-based geometric morphometrics），用于量化生物学中的体型与形态特征。诸多结构——实际上大多数动物——均为具有内部对称平面的两侧对称结构（亦称为对象对称（object symmetry））。在无需量化不对称变异的研究场景中，通常仅对单侧结构进行测量与分析。该研究范式已被数百项学术研究采用，其隐含假设为：另一侧的形态信息具有冗余性，仅分析单侧结构所得的结果，与对包含所有左右标志点（landmarks）的完整结构进行分析所得结果一致。然而，据笔者所知，这一核心假设的成立程度从未得到过系统性验证。本文依托两个示例数据集展开分析，结果表明：在宏观进化层面的研究中，单侧分析与双侧分析的结果一致性通常较高，但在微观进化层面则会显著降低，且形态特征的分析结果尤其容易出现偏差。此外，本文还将探讨其他可能影响分析结果的潜在因素，并提出一种简便易行的改进方案，以提升仅开展单侧分析的形态学研究的可视化效果与分析准确性。