Historical Leaf Trait Spectra from Minnesota Herbarium Specimens

201 herbarium specimens from 13 species provide reflectance spectra for leaf mass per area, nitrogen, and carbon content. Data spans specimens collected between 1876 and 2022 in the Minneapolis-St. Paul area. The dataset was created by Aaron Lee and the IHerbSpec organization.

Use Cases

• Train PLSR models to predict leaf mass per area (LMA) from vector-normalized adaxial reflectance spectra at 5nm resolution.
• Analyze temporal trends in foliar nitrogen content (N) using the specimen collection year metadata from 1876 to 2022.
• Compare spectral signatures between adaxial and abaxial leaf surfaces using the paired, processed spectra files.
• Assess model robustness by excluding NDVI (greenness) outliers or specimens based on age categories as outlined in the provided R scripts.

Strengths

• 201 specimens with empirical measurements for three key leaf traits (LMA, N, C).
• Spectral data is provided both as raw files and analysis-ready matrices processed to a final 5nm resolution.
• Includes R scripts for full analytical pipeline reproduction, from data extraction to figure generation.

Limitations

• Limited to 13 plant species from a single metropolitan area, reducing geographic and taxonomic generality.
• Sample size of 201 may be insufficient for training complex deep learning models without augmentation.
• Trait data is empirical but derived from preserved herbarium specimens, which may differ from fresh leaf measurements.

Provenance

Source

University of Minnesota Herbarium (MIN) specimens, processed by IHerbSpec.

Collection Method

Spectral data collected using an HR-1024i spectrometer with a modified leaf clip for flat specimens.

Time Range

1876 to 2022

Freshness

Last updated in April 2026.

Geography

Minneapolis-St. Paul metropolitan area, Minnesota, USA.