Semi-supervised determination of pseudocryptic morphotypes using observer-free characterizations of anatomical...

Area of Science:

• Biological shape analysis
• Geometric morphometrics
• 3D imaging in biology

Background:

• Accurate shape characterization is vital in biology.
• Automated 3D geometric morphometrics (auto3DGM) offers tools for shape analysis.
• auto3DGM's performance with pseudolandmark error needs further testing.

Purpose of the Study:

• To test auto3DGM sensitivity to variation and parameter settings.
• To evaluate auto3DGM using simulated and microCT mammal tooth crown data.
• To address critiques of auto3DGM regarding pseudolandmark placement error.

Main Methods:

• Utilized simulation and three microCT datasets of mammal tooth crowns.
• Varied datasets based on morphological differentiation and species.
• Tested auto3DGM sensitivity to pseudolandmark number and surface downsampling.

Main Results:

• Shape alignments are sensitive to the number of pseudolandmarks.
• Surface downsampling and morphological differentiation impact alignment repeatability.
• Critiques of auto3DGM stemmed from poorly parameterized alignments.

Conclusions:

• auto3DGM is reliable when pseudolandmark error is small relative to shape differences.
• Sample-specific sensitivity analyses are recommended for auto3DGM protocols.
• auto3DGM is a promising tool for morphometric studies across various biological scales.