How many landmarks are enough to characterize shape and size variation?

Area of Science:

• Evolutionary Biology
• Morphometrics
• Comparative Anatomy

Background:

• Accurate characterization of morphological variation is essential for reliable scientific conclusions.
• Landmark-based geometric morphometrics (GM) is widely used, but the adequacy of sampled landmarks is often unevaluated.
• A systematic method to assess landmark sampling fidelity in GM is lacking.

Purpose of the Study:

• To introduce LaSEC (Landmark Sampling Evaluation Curve), a computational tool for evaluating landmark sampling in GM.
• To provide a method for assessing the fidelity of morphological characterization based on landmark data.
• To guide users in optimizing landmark data collection for shape analysis.

Main Methods:

• LaSEC calculates the convergence of shape variation patterns from subsampled landmark data as sampling increases.
• The tool assesses how closely subsampled data reflect the full dataset's shape variation.
• Simulated shape data were used to evaluate the general properties of landmark data and LaSEC's performance.

Main Results:

• LaSEC identifies landmark under- and oversampling, assesses characterization robustness, and determines optimal landmark reduction.
• Results show that increasing landmark sampling generally improves the accuracy of morphological characterization, exhibiting statistical consistency.
• The number of landmarks required for adequate shape representation is dataset-dependent.

Conclusions:

• LaSEC offers a systematic approach to evaluate and refine landmark data collection in geometric morphometrics.
• Optimizing landmark sampling can reduce data collection costs and time, and potentially increase statistical power.
• This tool supports the accurate accumulation and analysis of morphological information, crucial for scientific advancement.