Connecting spectral and functional diversity at the leaf-level in Mediterranean herbaceous species: the DiverSpec monoculture experiment

Grasslands and tree-grass ecosystems play a fundamental role in the global carbon balance and the subsistence of the population in vulnerable regions. Protecting the entire range of ecosystem services provided by grasslands require assessing the influence of management and environmental drivers on these services and the role of biodiversity in their provision. Remote sensing offers tools that can help monitoring and better understand biodiversity in grasslands and its relationship with ecosystem function. The spectral diversity hypothesis suggests that spectral variations can be related to functional and phylogenetic diversity. Thus, different authors estimate foliar functional traits and validate the relationship between spectral optical properties of vegetation and functional diversity providing a powerful tool to understand the different roles species play in their environments. These studies mainly focus on forests, while functional characterization of grassland ecosystems is still limited (specially at leaf level) and key leaf traits, such as specific leaf area or cellulose and lignin content, remain underexplored. The phenology of grasslands has been also largely overlooked in biodiversity studies due to the challenges associated to field sampling. As a result, most datasets are collected only over short periods and do not represent the seasonality of the species and associated functional and spectral changes.

In this study, a monoculture experiment was implemented with 7 herbaceous species, including C3 and C4 grasses, legumes and forbs typical of Mediterranean grasslands to assess the capacity of hyperspectral data to detect intra- and inter-specific differences in foliar functional traits of pasture species at different phenological stages, and their plastic responses to water shortage. The experiment included 42 plots (1.5x1.5 m), with six replicates of every other species, organized in two blocks. Water regimes were manipulated to simulate typical versus water stress conditions. Leaf level reflectance was measured using a full range spectroradiometer ASD Fieldspec® 3 coupled with a plant probe and leaf clip with internal light source. Five regular measurements were carried out following the main phenological periods in the spring-summer growing season (April to June) 2022. Besides the reflectance data, key functional traits were also measured including leaf water content (LWC in g/cm²), leaf dry matter content (LDMC in %), specific leaf area (SLA in cm²/g), and chlorophyll and carotenoids concentrations (Cab and Car in mg/g). The potential of optical information to estimate foliar functional traits was explored using empirical models based on Partial Least Squares Regression (PLSR) techniques. Best fits (higher R² and lower normalized root mean squared error (nRMSE)) were achieved for LWC (R² = 0.94, nRMSE = 0.05) and Ca/Cb (R² = 0.89, nRMSE = 0.07), with slightly lower values for SLA (R² 0.71, nRMSE = 0.10). To investigate the seasonal dynamics of functional traits and spectral diversity, hierarchical clustering of the analyzed species based on observed and estimated foliar traits was calculated. Results revealed clear effects of phenology on the spectral diversity and the significant role of the LWC. This variable is not typically used for the functional characterization of herbaceous species.