Algorithmic approaches to aid species' delimitation in multidimensional morphospace

Area of Science:

• Paleontology
• Evolutionary Biology
• Taxonomy

Background:

• Species delimitation is a fundamental but contentious issue in biology.
• Morphological variability within samples can indicate single or multiple species.
• A statistical approach can objectively assess species boundaries.

Purpose of the Study:

• To present a novel statistical framework for species delimitation.
• To apply this framework to resolve taxonomic debates in fossil planktonic foraminifera.
• To provide statistical evidence for species delineation without pre-conceived hypotheses.

Main Methods:

• Utilizing statistical thresholds to analyze morphological variability.
• Employing robust covariance estimates and principal component analysis.
• Performing model-based clustering on dimension-reduced data.

Main Results:

• The framework successfully identified two distinct species of Eocene planktonic foraminifera, Turborotalia.
• Statistical evidence supports the existence of two species prior to the Eocene/Oligocene extinction event.
• The method avoids imposing pre-conceived patterns on species limit estimations.

Conclusions:

• Robust variance estimation and system-specific thresholds optimize species delineation.
• The framework enhances the objectivity of species delimitation in paleontology.
• This approach aids in resolving taxonomic disagreements by providing statistical evidence.