Estimating diversity via frequency ratios

Area of Science:

• Ecology
• Statistics
• Computational Biology

Background:

• Estimating total species diversity from limited sample data is challenging, especially with high latent diversity.
• Classical methods often rely on mixed Poisson models, which may not capture complex diversity patterns.

Purpose of the Study:

• To develop a novel statistical approach for estimating total population diversity from sample counts.
• To address limitations of existing models, particularly for datasets with high species richness.

Main Methods:

• Constructed a nonlinear regression model based on ratios of consecutive frequency counts.
• Utilized probability theory for distributions on integers.
• Applied the model to analyze high diversity datasets, including those from next-generation sequencing in microbial ecology.

Main Results:

• The proposed method provides accurate estimates of total diversity.
• The model demonstrates good data fits and reasonable standard errors.
• Outperformed existing competitor methods on a specific microbial ecology dataset.

Conclusions:

• This nonlinear regression approach offers a new, geometrically intuitive method for diversity estimation.
• The method is well-suited for analyzing complex, high-diversity ecological datasets.
• Represents a departure from traditional mixed Poisson models in diversity estimation.