A general method for parameter estimation in light-response models.

Area of Science:

• Plant physiology and biophysics
• Computational modeling and data analysis

Background:

• Accurate parameter estimation in nonlinear photosynthetic light response models is crucial for understanding plant productivity.
• Current methods often fail due to poor initial value selection, particularly with local optimization algorithms.
• A reliable, generalizable method for initial value selection is lacking.

Purpose of the Study:

• To develop a robust and generalizable method for parameter estimation in photosynthetic light response models.
• To overcome the challenge of initial value sensitivity inherent in local optimization techniques.
• To compare the performance of Differential Evolution (DE) against other global optimization algorithms.

Main Methods:

• Comparison of Levenberg-Marquardt algorithm with three global optimization algorithms.
• Development of a parameter estimation method utilizing Differential Evolution (DE).
• Application of the DE method to four distinct photosynthetic light response models.

Main Results:

• The DE method achieved high goodness-of-fit (R(2) > 0.98) across 42 datasets from 21 plant species.
• Robust parameter estimates were obtained using consistent initial values for all datasets.
• DE effectively resolved the hyper initial-value sensitivity issue associated with local optimization.

Conclusions:

• Differential Evolution (DE) provides an efficient and reliable solution for parameter estimation in photosynthetic light response models.
• The DE method eliminates the need to carefully select initial values, simplifying model application.
• This approach enhances the fitting of light-response curves for diverse plant species.