Abstract
Plant pathogens are increasingly compromising forest health, with impacts to the ecological, economic, and cultural goods and services these global forests provide. One response to these threats is the identification of disease resistance in host trees, which with conventional methods can take years or even decades to achieve. Remote sensing methods have accelerated host resistance identification in agricultural crops and for a select few forest tree species, but applications are rare. Ceratocystis wilt of 'ōhi'a, caused by the fungal pathogen Ceratocystis lukuohia has been killing large numbers of the native Hawaiian tree, Metrosideros polymorpha or 'Ōhi'a, Hawaii's most common native tree and a biocultural keystone species. Here, we assessed whether resistance to C. lukuohia is detectable in leaf-level reflectance spectra (400-2500 nm) and used chemometric conversion equations to understand changes in leaf chemical traits of the plants as indicators of wilt symptom progression. We collected leaf reflectance data prior to artificially inoculating 2-3-year-old M. polymorpha clones with C. lukuohia. Plants were rated 3x a week for foliar wilt symptom development and leaf spectra data collected at 2 to 4-day intervals for 120 days following inoculation. We applied principal component analysis (PCA) to the pre-inoculation spectra, with plants grouped according to site of origin and subtaxon, and two-way analysis of variance to assess whether each principal component separated individuals based on their disease severity ratings. We identified seven leaf traits that changed in susceptible plants following inoculation (tannins, chlorophyll a+b, NSC, total C, leaf water, phenols, and cellulose) and leaf chemistries that differed between resistant and early-stage susceptible plants, most notably chlorophyll a+b and cellulose. Further, disease resistance was found to be detectable in the reflectance data, indicating that remote sensing work could expedite Ceratocystis wilt of 'ōhi'a resistance screenings.