Hyperspectral monitoring of English oak responses to drought and ozone

The Mediterranean basin is considered a global biodiversity hotspot, but Mediterranean plants are threatened by natural and anthropogenic factors such as drought and tropospheric ozone (O3), which are expected to be harsher in the near future. Advancements in techniques to rapidly and non-destructively detect and monitor the impact(s) of harsh environmental conditions on vegetation are thus required to ensure the maintenance of ecosystem services in these environments. This study examined the capability of full-range (350-2500 nm) reflectance spectroscopy to characterize responses of two-year old English oak (Quercus robur) saplings firstly subjected to drought for one month, and then to both drought and O3 (120 ppb of O3, 8 h day-1) for other 10 consecutive days. Analyzing spectral signatures collected from asymptomatic leaves, we accurately discriminated plants exposed to different drought and O3 conditions, as well as to different combinations of the two stressors (overall accuracy: 92, 75 and 73%, respectively). Furthermore, using a partial least squares regression (PLSR) approach, we constructed predictive spectral models to estimate from spectra an array of leaf traits commonly investigated to elucidate physiological (e.g., gas exchange, chlorophyll a fluorescence), water status (e.g., water and osmotic potentials, relative water content) and morphological responses (e.g., leaf thickness) of plants to stress. These spectral models were developed using data collected on both young saplings and mature trees. Most traits were relatively well predicted by spectral models (model goodness-of-fit for validation, R2: 0.60-0.86). Finally, variations of spectral vegetation indices and leaf traits derived from spectra by developed PLSR-models suggested a good tolerance of English oak to both drought and O3, even when applied together, proposing this species as a good candidate to face future predicted challenges related with climate change. Overall, the present study highlights the potential of hyperspectral data to improve the feasibility of assessing tree disturbances as a result of environmental stress in a timely and cost-effective manner.