Predicting leaf traits across functional groups using reflectance spectroscopy.

Area of Science:

• Plant Ecology
• Remote Sensing
• Spectroscopy

Background:

• Plant functional traits are crucial for understanding plant responses to and impacts on their environment.
• Reflectance spectroscopy offers a rapid, non-destructive method for estimating leaf traits.
• The generalizability of trait-spectra models across diverse plant functional groups and ecosystems requires further investigation.

Purpose of the Study:

• To develop and validate general trait-spectra models for estimating multiple leaf functional traits.
• To assess the accuracy and generalizability of these models across different plant functional groups and ecosystems.
• To evaluate the potential of spectroscopy for global plant function monitoring.

Main Methods:

• Collected leaf spectral data and measured 22 structural and chemical traits from nearly 2000 samples across 103 species.
• Employed partial least-squares regression (PLSR) to build empirical models linking leaf spectra to traits.
• Validated model performance on external datasets to assess generalizability.

Main Results:

• PLSR models achieved high accuracy for predicting leaf mass per area (LMA) and leaf dry matter content (LDMC) (R² > 0.85).
• Models for chemical traits (pigments, carbon, nutrients) demonstrated intermediate accuracy (R² = 0.55–0.85), with micronutrients showing lower precision.
• External validation showed good performance for LMA and LDMC, but reduced accuracy for carbon fractions.

Conclusions:

• Developed robust, generalizable models for estimating key plant functional traits from leaf spectra.
• Spectroscopy provides a powerful tool for rapid and reliable assessment of plant traits.
• Findings support the expanded use of spectroscopy for monitoring plant function globally.