Assessing the performance of DNA barcoding using posterior predictive simulations.

Area of Science:

• Biodiversity science
• Ecology
• Evolutionary biology
• Systematics
• Conservation biology

Background:

• Accurate biodiversity estimates are crucial for ecology, evolution, systematics, and conservation.
• DNA barcoding and statistical models are proposed tools to address gaps in biodiversity knowledge.
• The reliability of DNA barcoding for biodiversity estimates hinges on the accuracy of underlying evolutionary process models.

Purpose of the Study:

• To assess DNA barcoding performance using commonly used substitution models.
• To evaluate the impact of model choice on biodiversity estimates.
• To determine the statistical justification for using the Kimura 2-parameter model in DNA barcoding.

Main Methods:

• Utilized data from the Barcode of Life Database.
• Employed posterior predictive simulations to test statistical models.
• Assessed model performance across various taxonomic groups, clustering algorithms, genetic divergence thresholds, and substitution models.

Main Results:

• DNA barcoding success varied substantially across different DNA substitution models.
• Model choice influenced the number of identified operational taxonomic units by approximately 4-31%.
• The common practice of a priori assuming the Kimura 2-parameter model was found to be statistically unjustified.

Conclusions:

• Model selection and adequacy are critical for accurate biodiversity quantification using DNA barcoding.
• The choice of substitution model significantly impacts biodiversity estimates.
• Avoid a priori assumptions of specific models like the Kimura 2-parameter model in DNA barcoding studies.