Handheld Fourier-transform infrared spectroscopy as a portable tool for tree health

This dataset contains FT-IR spectral data and trained machine-learning models used to evaluate handheld mid-infrared spectroscopy for phenotyping Pinus sylvestris and Fraxinus excelsior. It includes pre-processed spectra, metadata, and the sPLS-DA and SVM models used in the manuscript analyses. These files allow full reproduction of classification results and enable independent testing of handheld FT-IR for forest-health applications. This repository contains all data and trained model objects associated with the manuscript: “Handheld FT-IR spectroscopy for rapid phenotyping of forest trees: evaluating infection status and genotypic variation in Pinus sylvestris and Fraxinus excelsior.” The deposit includes: 1. Raw and processed FT-IR spectral data Pine_FTIR_avr_2der_TRIM.csv – averaged and pre-processed mid-infrared spectra for Pinus sylvestris needles and stem phloem at 0, 14, and 21 days post inoculation with Diplodia sapinea. Ash_FTIR_avr_2der_TRIM.csv – averaged and pre-processed spectra for Fraxinus excelsior leaf, shoot, and stem tissues from multiple genotypes and field sites. Each file contains: Sample metadata (tree ID, genotype, tissue type, location, inoculation status/date) Wavenumber-intensity values (trimmed and 2nd-derivative transformed) These datasets correspond exactly to the versions analyzed in the manuscript. 2. Trained machine-learning model files (“model weights”), stored as .rds objects (R’s native serialized format) for full reproducibility. 3. Minimal scripts for model testing (see readme file).