Date of Award

Fall 12-2021

Level of Access Assigned by Author

Open-Access Thesis

Degree Name

Master of Science (MS)

Department

Botany and Plant Pathology

Advisor

José Eduardo Meireles

Second Committee Member

Peter R. Nelson

Third Committee Member

Daniel Stanton

Additional Committee Members

Jacquelyn Gill

Brian McGill

Abstract

Biodiversity is changing and it is imperative that we continually assess it in order to preserve ecological services that we rely on. Spectral platforms are increasingly being used to assess biodiversity due to the fact that light reflected from an organism’s surface carries much of information about that organism. Despite the promise spectroscopy shows, two gaps in our knowledge remain. First, we do not know how well reflectance spectra can be used to estimate fine-scale diversity – intraspecific genetic and phenotypic diversity – that is fundamental to ecological and evolutionary processes. Second, spectral libraries, used to construct models to estimate diversity, have largely been built from plant spectra and have neglected other ecologically important organisms such as lichens. To investigate the first problem, my colleagues and I tested the utility of reflectance spectra for distinguishing genomically defined populations. We collected spectra (400–2400 nm) and samples from co-occurring Dryas alaskensis, Dryas ajanensis, and hybrid individuals from six different mountaintops in the interior of Alaska, United States. We used partial least squares discriminant analysis (PLS-DA) to classify leaf reflectance spectra into six populations defined by STRUCTURE and PCA analyses using genomic data. We also estimated the phylogenetic signal carried by the spectra, and we used PLS beta regression to estimate the proportion of ancestry for each individual from the reflectance spectra. We found that the two species and their six populations could be distinguished with 99.7% and 98.9% overall accuracy, respectively. A significant phylogenetic signal was found for all regions of the spectrum, and the model for estimating the proportion of ancestry explained 91% of the variation with an RMSE of 0.13. Hybrids were classified with 80% accuracy, and this is thought to be due to a lack of strong trait correlations. These findings suggest that fine-scale diversity can be retrieved from reflectance spectra and this should be considered in future spectrally-based biodiversity assessments. To address the second problem, I investigated whether herbarium specimens would be valuable for building a spectral library for lichens. Specifically, I investigated whether lichen specimens were altered by the long-term desiccation inherent with herbarium storage and if that influenced the classification of herbarium specimens. I used a spectral dataset of 30 lichen species that covered an age range of 126 years, and used linear mixed-effects models and PLS-DA to determine 1) how reflectance changed with age, and 2) the influence of age on classification accuracy. I found that the reflectance for wavelengths between 700 and 1900 nm decreased by less than 0.2% reflectance per year, but wavelengths outside this range did not clearly respond to aging. This implies a gradual change in thallus structure over time in herbarium storage. Species, families, orders, and classes were classified with 77.0 to 94.5% accuracy, and these models were only marginally influenced by specimen age. These results indicate that lichen specimens do change over time, but these changes do not negate their utility for building spectral libraries for lichens.

Download

Included in

Botany Commons

Share