Mapping water patterns during dehydration of nonvascular epiphytic communities through NIR hyperspectral imaging

Description

The present work is aimed at developing water maps for understanding how a micro-ecosystem composed by nonvascular epiphytic communities (mainly, lichens and bryophytes), manages water and, consequently, how the thalli can prolong water availability depending on the morphology of the species and the composition of the community itself. To achieve this aim, the water status of bark, lichens (crustose and foliose) and bryophytes was monitored by means of near-infrared (NIR) hyperspectral imaging (HSI), working in the spectral range from 1000 to 2500 nm (SWIR3 – Specim, Finland). Pure communities, composed mainly by a single organism, were analysed alongside more complex samples containing two or three lifeform coexisting in the same ecosystem, for a total of 32 samples. All the micro-ecosystems were naturally supported by chestnut bark. Samples of non-colonised (or as less colonised as possible) bark were also analysed. The samples were first submitted to a full hydration protocol, then let dry at room temperature (19°C) and humidity (41%) for about 12 h. During the dehydration, NIR-HS images were acquired every 45 minutes, and then twice more to get to full dehydration, getting 17 sampling points and 544 images. Before image acquisition at each sampling point, the samples were weighed, and the weight was used to calculate their water. Exploratory data analysis was performed on the images by means of principal component analysis (PCA), with the aim of comparing the drying trajectory in the orthogonal space defined by the lowest-order components with the dehydration curve. The study of these correspondences on pure communities opens the possibility to understand the water management of each organism and then to use such information in the study of more complex microclimates (binary and ternary samples). In practice this is done projecting the images in the PCA space defined by the pure samples and then producing false-RGB score maps. After the exploratory step, a supervised strategy is proposed, with the aim of extracting not only the water pattern inside the microclimate but also a numerical parameter expressing the water content in the different regions of interest inside a single map. This approach enabled to observe distinct behaviours between different types of communities inside a single sample. The study of these communities and their relative behaviour in water stress condition plays a crucial role in understanding how these organisms are facing the climate change and the subsequent water scarcity or overabundance.

Additional details