Tree growth inference and prediction from diameter censuses and ring widths.

Area of Science:

• Ecology
• Forestry
• Statistical Modeling

Background:

• Tree growth estimation relies on limited data (diameter, ring width, increments).
• Existing methods struggle to integrate diverse data types and account for observation errors, data sparsity, shared variance, and non-independent growth rates.
• Previous approaches often yield impossible negative growth estimates.

Purpose of the Study:

• To develop a hierarchical Bayes state space model for tree growth estimation.
• To formally integrate diverse data sources (e.g., tree rings, census data) while addressing estimation challenges.
• To enable accurate inference and prediction of tree growth, even with sparse and error-prone data.

Main Methods:

• Developed a hierarchical Bayes state space model.
• Incorporated multiple sources of observation error and data sparsity.
• Accounted for shared population variance and non-independent tree growth within stands.
• Ensured non-negative growth estimates.

Main Results:

• The model successfully integrates tree-ring and census data from multiple species and stands.
• It provides formal inference consistent with data and non-negative growth assumptions.
• The approach allows for uncertainty incorporation in predictions and gap-filling for past/future growth.

Conclusions:

• The hierarchical Bayes state space model offers a robust solution for tree growth estimation.
• It effectively addresses limitations of traditional methods, improving accuracy and reliability.
• This model enhances our ability to predict forest dynamics under various conditions.