Introduction
The rate of photosynthesis is dynamically regulated by stomatal responses to environmental factors (e.g., temperature, water availability) in concert with leaf biochemical capacities. In doing so, stomata influence the marginal water cost of carbon acquisition at the leaf scale and more broadly affect the coupling between carbon and water cycles, which is especially important in light of a changing climate (Damour et al., 2010; Duursma et al., 2013; Gimenoet al., 2016; Héroult et al., 2013). However, it is not well understood whether the water-carbon trade-off will shift towards a more profligate or more conservative water use strategy in a changing climate. To address this knowledge gap, we used 11 years of data from an ecologically realistic long-term climate change experiment to test whether and how the trade-off between water loss and carbon gain changes with modest experimental warming and rainfall reduction in 21 tree species at the boreal-temperate forest ecotone.
One of the long-standing hypotheses about stomata is that their behavior follows optimization theories (Cowan, Farqhuar, 1977; Medlyn et al., 2011; Manzoni et al., 2013; Wolf et al., 2016) with stomatal behavior often modeled as a gain-cost trade-off that maximizes carbon gain through variation of stomatal conductance in response to environmental constraints (e.g., water availability and temperature). Cowan and Farquhar (1977) proposed a mechanistic model where the role of stomata is to maximize carbon acquisition (A ) at the lowest water loss through transpiration (E ), described as a marginal water cost of carbon gain (λ). Therefore, optimal stomatal behavior minimizes the integrated sum of the following expression – that effectively defines the marginal water cost of carbon gain – and can be written as:
𝐸 - 𝜆𝐴 (1)
where:
A – photosynthesis,
E – transpiration,
λ – is a parameter representing the marginal water cost of carbon gain.
However, fully mechanistic models like the one proposed by Cowan and Farqhuar (1977) (equation 1) are difficult to parameterize with typically measured field data, leading to models using proxies to represent λ. Following many important predecessors (e.g., Ball, Woodrow & Berry et al., 1987; Leuning, 1995, and others), Medlynet al., (2011) developed the Unified Stomatal Optimization model (USO), that describes stomatal conductance as a function of carbon assimilation and environmental conditions (\(\frac{A}{\left(C_{a}\sqrt{D}\right)}\) where: A – is net assimilation rate, Ca – is atmospheric CO2 concentration at the leaf surface, D – is vapor pressure deficit (kPa) at the leaf surface). As derived by Medlynet al., (2011) the slope (g1 ) of the USO model gains biological meaning by combining equations of standard leaf diffusion with optimum leaf internal CO2 concentration (Ci) to link the g1 parameter with λ (Arneth et al., 2002); for detailed description of the model see Medlyn et al., 2011 and supplement). The slope of the USO model (g1 parameter) is directly proportional to the combination of λ and CO2 compensation point (Γ*):
\(g_{1}\propto\sqrt{\Gamma^{*}*\lambda}\) (2)
where:
Γ* – is CO2 compensation point,
λ – is marginal water cost of carbon gain.
This linkage allows interpretation of the slope parameterg1 , where low values represent conservative water use while higher g1 indicates more profligate use, and the development of testable hypotheses, including about the response of stomatal conductance to novel environmental conditions such as elevated temperatures and reduced water availability.
Thus, the g1 parameter should increase with λ and Γ*, assuming that Ca is much larger than Γ* (Ehleringer, 2005) and that stomatal behavior optimizes for RuBP (Ribulose 1,5-biphosphate) regeneration limitation but not for Rubisco limitation of photosynthesis (Outlaw et al.,1979; Outlaw & De Vlieghere-He, 2001; Shimazaki, 1989). Becauseg1 is proportional to the\(\sqrt{\Gamma^{*}*\lambda}\) term it can be assumed that it will be sensitive to water availability, temperature, and CO2concentration, and will vary as those do. Thus, it is expected thatg1 will decrease with decreasing water availability, and because Γ* is exponentially dependent on temperature (Bernacchi et al., 2001),g1 should increase with increasing growth temperature.
Despite theoretical predictions summarized above, and many studies of photosynthesis and stomatal conductance in relation to climate, empirical evidence about the trade-off between carbon gain and water loss remain limited (Lin et al., 2015; Medlyn et al.,2016) especially concerning individual species representing different biomes, plant types, and responses to potential future climates. Moreover, while there has been considerable research on the impacts of single climatic drivers on stomatal behavior, we lack research on multiple climatic drivers and multiple species (Atkinson & Urwin, 2012; Stevens et al., 2021). As a result, the effects of climate warming and water availability are highly uncertain and poorly represented in many models (from leaf to global scale) and in particular are not well parameterized in terms of drought sensitivity. By changing evaporative demand and/or soil moisture, both temperature and rainfall variation might change the optimal water cost and thus stomatal conductance and net photosynthetic rates. Additionally, given that species differ in their adaptations and sensitivity to both warm temperatures and limited water availability, we might also expect plants to differ systematically in terms of their stomatal behavior, water use efficiency (WUE), and how they modulate these as air and soil moisture conditions vary as pushed by a changing climate (Medlyn et al.,2011).
To address this knowledge gap, we evaluate the g1parameter for a suite of temperate and boreal tree species grown in a realistic experimental setting that mimics climate change drivers (i.e., warming and rainfall reduction). Our main goals were to explore whether and how (i) stomatal behavior changes (e.g., plants decreaseg1 indicating more conservative water use) in response to climate warming or soil water deficits induced by experimental treatments; (ii) stomatal behavior varies with species identity, drought tolerance, and biome association; and to determine (iii) whether there are generalizable patterns across species, their associations (e.g., biome) and environmental conditions: i.e. do species differ in response to climate treatments, and are responses to warming and rainfall reduction additive or interactive? To achieve those goals, we addressed the following issues and hypotheses.
First, consistent with optimization theory that predicts a decrease of λ with declining water availability, we hypothesize thatg1 (which is proportional to λ, see equation 2) will decrease as soil moisture declines (e.g., Lu et al., 2016). This leads to H1: g1 decreases with reduced rainfall for all species .
Second, equation 2 suggests that g1 will increase with warming because; i) it is proportional to Γ*which is dependent on temperature (Bernacchi et al., 2001), ii) temperature induced changes in wood density will affect hydraulic conductivity (Héroult et al., 2013; McCulloh et al.,2016), and iii) increasing temperature lowers the viscosity of water making it cheaper to transport (Yamamoto, 1995). However, we hypothesize that soil drying induced by warming treatments will causeg1 to decrease. Thus, the ultimate influence of warming will depend upon a balance between the direct influences of temperature that should increase g1 and the indirect influence of temperature on soil moisture that should decreaseg1 . We expect the direct warming effects ong1 to be modest at best, and therefore the response of g1 to warming to be dominated by soil moisture (Reich et al., 2018). This leads to H2: g1 will decrease with climate warming due to soil moisture reduction induced by elevated temperature .
The interactions of warming and rainfall reduction do not easily lend themselves to a simple hypothesis, due to the complexity of both direct and indirect effects of elevated temperature on factors that might influence g1 (such as leaf temperature and soil moisture), and uncertainty about whether those effects are contingent on rainfall levels. However, because the effects of warming and reduced rainfall do not have a consistent interaction on VWC (Volumetric Water Content in soil) at our study sites (data not shown), we hypothesize