Remote sensing reveals inter- and intraspecific variation in riparian cottonwood (Populus spp) response to drought

1 INTRODUCTION

Climate change exacerbates drought conditions in many regions (Dai, 2011; Stocker et al., 2014), increasing vegetation stress, plant mortality and biodiversity loss while disrupting critical mutualisms and ecosystem stability (Allen et al., 2010; Au et al., 2023; Steinkamp & Hickler, 2015; Stone et al., 2018). Prolonged and severe drought conditions threaten vital ecosystem services (Breshears et al., 2011; Wolf & Paul-Limoges, 2023), yet forests have demonstrated resilience and adaptive capacity to water stress (Amlin & Rood, 2003; Anderegg et al., 2018; Phelan et al., 2022). This resilience can be driven by genotype-by-environment (G × E) interactions, as both intraspecific and interspecific variations shape ecosystem responses to drought (Anderegg et al., 2018; Grossiord, 2020; Rodríguez-Alarcón et al., 2022). Remote sensing is a powerful, non-destructive tool for examining G × E interactions at multiple scales, capturing forest responses to water stress across species, populations and landscapes.

Spectroscopy and thermal imaging are particularly effective for studying G × E interactions in the context of drought, as they capture physiological responses to water stress across spatial and temporal scales (Cotrozzi et al., 2017; Le et al., 2023; Li et al., 2023; Sapes et al., 2024). Spectroscopy, which captures data across the visible through shortwave infrared (VSWR; 350–2500 nm) spectrum at short wavelength intervals (e.g. 1–10 nm), allows us to detect changes in plant traits indicative of water stress (e.g. rehydration capacity, leaf water potential, relative water content, electrolyte leakage) before they are visually apparent (Cotrozzi et al., 2017; Mohd Asaari et al., 2022; Sapes et al., 2024). Water stress, quantified as canopy water content, has been estimated across large spatial scales using imaging spectroscopy (Asner et al., 2016). These data further capture changes in other leaf traits (e.g. leaf mass per area, chlorophyll content; Asner & Martin, 2016), providing a non-destructive measure of plant phenotypes at both the leaf and landscape levels. While the entire VSWIR spectrum is used to capture plant physiological responses, specific regions contribute distinct information: the visible range (400–700 nm) largely reflects photosynthetic pigments like chlorophyll, the near-infrared (NIR; 750–1300 nm) detects variation in cell structure and water content, and the shortwave infrared (SWIR; 1300–2500 nm) is often indicative of non-photosynthetic plant traits such as lignins and tannins (Asner & Martin, 2016). Furthermore, recent studies by Corbin et al. (2025) show that spectroscopy is adept at discriminating among populations and genotypes and detecting G × E interactions of P. fremontii, one of the species used in the present study. By detecting drought stress and phenotypic responses across environments, spectroscopy captures plant water-use strategies and adaptive capacities, forming a foundation to advance our understanding of local adaptations, community structure, phytochemistry and stress responses as well as for scaling these analyses to broader spatial extents through complementary remote sensing methods such as thermal imaging.

Thermal imaging is another important remote sensing tool for investigating variation among water-use strategies, plant water status and drought tolerance at large spatial scales (Fuchs, 1990; Scherrer et al., 2011). Leaf and canopy temperatures, mediated by transpiration via evaporative cooling (latent heat loss), rise when water availability is limited, making thermal imaging a reliable indicator of water stress. This approach has been successfully applied in plant and ecosystem sciences using platforms such as towers, unmanned aerial vehicles (UAVs) and satellites (Farella et al., 2022). By capturing G × E interactions at landscape scales, thermal imaging and spectroscopy offer a means of understanding ecosystem responses as climate change intensifies drought across many regions.

Given the growing importance of understanding G × E interactions in the context of climate change, cottonwoods offer ideal case studies for investigating water stress in trees due to their significant intra- and interspecific variation, which mediates their response to environmental stressors (Blasini et al., 2021, 2022; Bothwell et al., 2023; Hultine et al., 2020; Kaluthota et al., 2015; Moran et al., 2023; Woolbright et al., 2008). Two cottonwood species, (Populus fremontii Wats. and P. angustifolia James) and their naturally occurring hybrids coinhabit riparian zones throughout the southwestern U.S. but also differ in their hydraulic traits. Populus fremontii (Fremont cottonwood) has higher canopy conductance (Fischer et al., 2004) and less stomatal sensitivity to cumulative vapour pressure deficit over the growing season than its sister species, P. angustifolia (narrowleaf cottonwood; Guo et al., 2022). Further, hybrids between P. fremontii and P. angustifolia display distinct genetic and ecological characteristics, such as the formation of hybrid zones that consist of F₁ hybrids and backcross genotypes with the potential for increased tolerance to drought (Hultine et al., 2020; Martinsen et al., 2001; Whitham et al., 1999). Hybrids are of special interest as their unique genetic combinations (F₁s to backcrosses; Hersch-Green et al., 2014) provide targets for selection and the potential for evolution in response to climate change (see review by Peñalba et al., 2024). Drought resilience in both Populus species is a pressing conservation concern as they constitute foundation species (Hultine et al., 2020) that have declined due, in part, to water limitations, leading to the decline of riparian ecosystems in the American southwest (Braatne et al., 1996; Hultine et al., 2020; Stromberg, 2001; Stromberg et al., 1996).

Intraspecific differences—shaped largely by past climatic conditions—also play a key role in a species' ability to tolerate water stress (Gazol et al., 2023; González de Andrés et al., 2021; Jung et al., 2014; Luo et al., 2023). In P. fremontii, intraspecific variation is pronounced, with populations exhibiting distinct physiological adaptations to temperature and water availability, consistent with the recognition of distinct ecotypes across its range (Bothwell et al., 2023; Hultine et al., 2020; Ikeda et al., 2017). For example, warm-adapted P. fremontii populations use high transpiration rates during heat waves to keep leaf temperatures below thermal thresholds (Blasini et al., 2021; Moran et al., 2023; Posch et al., 2024). However, this greater capacity for leaf cooling comes at the cost of operating closer to hydraulic failure thresholds than cool-adapted populations (Blasini et al., 2021; Posch et al., 2024), given their high susceptibility to drought-induced xylem cavitation (Leffler et al., 2000; Tyree et al., 1994).

We investigated G × E interactions using leaf-level VSWIR spectroscopy and canopy-level UAV thermal data by exploring intra- and interspecific drought responses in cottonwood (Populus spp.). First, we explored interspecific drought responses of three Populus cross types, P. fremontii, P. angustifolia and their hybrids, using leaf VSWIR reflectance data collected during two greenhouse drought experiments. We focused on quantifying their spectral shifts and phenotypic convergence under environmental stress, hypothesizing that, while the spectral space of the three cross types will shift as a result of drought, their spectra will be distinct regardless of drought status. We additionally assessed intraspecific G × E interactions in P. fremontii by evaluating how variations in source population mean annual temperature (MAT) impacted drought responses in leaf-level VSWIR and canopy-level UAV thermal data from two common garden experiments. We hypothesized that intraspecific spectral and thermal responses to drought will reflect historic climate adaptations to heat and drought, and warm-adapted populations will exhibit lower leaf temperatures pre-drought but will be more impacted by drought than cool-adapted populations. By investigating the drought response of foundation tree species using remote sensing, we sought to explore and demonstrate remote sensing as a tool for quantifying and monitoring intra- and interspecific variation in response to environmental stressors at local to landscape scales.