Replication Data for: \"Seeing herbaria in a new light: leaf reflectance spectroscopy unlocks trait and classification modeling in plant biodiversity collections\"

This dataset contains hyperspectral reflectance files, sample metadata, and combined files containing processed spectral data, trait data, and metadata derived from specimens in the Harvard University Herbaria. These represent 25 plant species sampled from New England over the past 179 years. We also include the processed spectra and leaf trait data from pressed leaf dataset of Kothari et al. 2023 (see Related Publication). It supports reproducible herbarium-based spectroscopy research as documented in the associated study. All code and analytical workflows used to generate and analyze these data are available at: https://github.com/Erythroxylum/herbarium-spectra/ Most metadata fields are defined and implemented according to the IHerbSpec Protocol v1.0. Accessible at: https://doi.org/10.5281/zenodo.15849668 Several additional metadata and trait variables used in this dataset are not defined in the IHerbSpec Protocol v1.0, but are described in Table 3 of the associated article and summarized here: Green Index: Calculated as (Reflectance550nm − Reflectance690nm) / (Reflectance550nm + Reflectance690nm) Age: Years since the specimen was collected Day of Year: Julian day of collection Leaf Mass per Area: Measured in kg·m−2 Nearest Taxon Distance: Estimated age (in millions of years) of the most recent common ancestor shared between the predicted taxon and the nearest sampled species File descriptions: DMWhiteHUHspec1_sp25leaf560_noResample_350-2500.csv: For new spectral analyses starting with unprocessed spectra, this is a spreadsheet containing sample metadata, leaf trait data, and unprocessed reflectance spectra across the full wavelength range (338.6-2514.6 nm). Each row corresponds to a scan from a single leaf surface of a herbarium specimen. DMWhiteHUHspec1_processed_spectra_files.tar.gz: For replication of the analyses in the NPH article, this is a compressed archive containing six .csv data files that each combine sample metadata, leaf trait data, and processed spectral data. Three files pertain to the Harvard University Herbaria samples, and the other three are the pressed leaf spectral data from Kothari et al. (2023) - BUT UNAVERAGED. For each data source, the three files represent: reflectance spectra trimmed and resampled at 5 nm resolution, continuous wavelet transformed spectra, and vector normalized spectra together with the same sample metadata and leaf trait data. DMWhiteHUHspec1_raw_spectra_files.tar.gz: For new analyses beginning with unprocessed spectral files, this is an archive of 1,690 original spectral .sig files recorded using the SVD HR-1024i spectroradiometer. Filenames follow IHerbSpec Protocol v1.0 conventions and correspond to the below DMWhiteHUHspec1_sp25leaf560_metadata.csv file. DMWhiteHUHspec1_sp25leaf560_metadata.csv: Metadata-only CSV listing filenames along with Harvard University Herbaria specimen and measurement details for each unprocessed spectra file included in DMWhiteHUHspec1_raw_spectra_files.tar.gz.