**1. Introduction**

Plant growth and development involve numerous biochemical reactions that are sensitive to temperature [1]. At the leaf level, high temperatures may lead to a decrease in photosynthesis and an increase in respiration, thus reducing ovetrall net carbon dioxide (CO2) assimilation rate [2]. Under extreme conditions, plant thermotolerance might be exceeded, leading to permanent damage to the photosynthetic machinery [3]. Leaf thermal damage occurs when leaf temperature (TL) exceeds a critical temperature (TCRIT), but the damage becomes permanent if TL exceeds an even higher thermal threshold, often termed 'maximum temperature' [3–7].

Leaf temperature is defined by the interplay between plant traits (chiefly, leaf size and shape [8], albedo [9] and stomatal conductance [10]) and environmental conditions (irradiance, air temperature and humidity, plant water availability). When plants are well-watered and the canopy is well-coupled with the surrounding atmosphere, air temperature (TA) and TL are generally well correlated, although thermal regulation can result in leaves being cooler than air in warm conditions and warmer otherwise [11]. Low wind speed and plant water shortage can lead to TL substantially exceeding TA [12–14], by reducing the leaf evaporative cooling.

Climate change projections suggest a global increase in the frequency of warm days and nights, heat waves and, in some locations, dry spells [15]. Boreal forests are warming up twice as fast as other ecosystems. As a result of higher temperatures and, in some regions, reduced summer precipitation [15], they will be subject to more frequent periods of low water availability. Warming, together with low water, is already negatively affecting forest productivity [16,17] and threatening the survival of sensitive species [18]. This is especially problematic in the boreal region considering the role of boreal forests in the global carbon (C) cycle [19,20] and as a biomass source in climate change mitigation policies [21]. However, while the joint effects of high air temperatures and low water availability and the occurrence of thermal damage and/or limitation of assimilation rates have been extensively studied in arid and semi-arid environments [22,23], the potential occurrence of such conditions has been mostly overlooked in more mesic environments and boreal forests [24]. To better understand the joint effects of high temperatures and low water availability is essential in order to identify the expected frequency of damaging conditions the potential extent of damage, and possible steps towards preparing boreal forests to future growing conditions.

One potential and still partially unexplored avenue for adaptation to climate change is the identification of combinations of plant traits that can support carbon uptake and reduce the risk of thermal damage in a changing climate [25]. In particular, the role of traits leading to high temperature tolerance has been little explored over boreal regions [26]. Similarly, the link between plant traits, water use and water stress has not been fully resolved in boreal forests [27]. Limited information and understanding may be partially due to the fact that measurements of comprehensive sets of traits and long-term monitoring of forest dynamics are expensive and time consuming, constraining the traits considered [27]. Mechanistic models offer a powerful tool to overcome these limitations, allowing the exploration of a wide set of traits and environmental conditions, including the expected warmer and drier conditions.

Here, a mechanistic model describing energy and mass exchanges among the soil, plant and atmosphere [28,29] is applied to explore the role of plant traits and current and future environmental conditions on leaf temperature and net CO2 assimilation, with a focus on boreal forests. Specifically, we answer three questions, pertaining traits, growing conditions and their joint effects respectively: (i) Which are the dominant plant traits in regulating leaf temperature and CO2 assimilation? (ii) How do environmental conditions (e.g., air temperature and precipitation timing) interact in determining the occurrence of thermal damage and reduced CO2 assimilation? (iii) Which are the most suitable trait combinations for sustained CO2 assimilation and thermal damage avoidance? How do they differ between current and future, warmer and drier conditions? By answering these questions, this study identifies the key plant traits along the thermal safety—C fixation tradeoff, under current and future climates. It also suggests optimal trait combinations for thermal safety and productivity to cope with future climates. Therefore, this study contributes to disentangle the traits—growing conditions nexus, including the joint effect of increasing temperatures and more frequent lack of water in boreal forests.

#### **2. Materials and Methods**

#### *2.1. APES-Atmosphere-Plant Exchange Simulator Model*

We use the model APES (Atmosphere-Plant Exchange Simulator), previously developed, calibrated and validated in boreal forests. APES is a process-based one-dimensional multilayer, multi-species forest canopy-soil model, designed especially to account for the vertical structure and functional diversity. The model mechanistically solves the coupled energy, water and C cycles in the soil-vegetation-atmosphere system, using physical and physiological constraints. Only the modeling aspects most relevant for the purposes of this work, i.e., those related to the determination of leaf temperature and net CO2 assimilation rate, are briefly described here and in the Supplementary Materials. A complete description of this model and its validation, as well as some applications, can be found in Launiainen et al. [28,29].

In APES, the leaf temperature, TL, is calculated separately for sunlit and shaded leaves within each canopy layer, by solving the coupled energy balance and net CO2 assimilation rate (Anet) (see the Supplementary Materials for the equations).

Anet is obtained based on the Farquhar model, as the minimum of Rubisco-limited and light-limited rate, which in turn depend on the maximum carboxylation rate VCMAX and the maximum electron transport rate JMAX, respectively [30]. The temperature responses of both VCMAX and JMAX are as in Medlyn et al. [31]. Additionally, leaf respiration increases with TL, so that Anet reaches its maximum at an intermediate temperature (the optimal temperature for net CO2 assimilation). The parameters VCMAX, JMAX and Rd at reference temperature are affected by the leaf water potential (ψ*L*), following Kellomaki and Wang [32]. The stomatal conductance is computed by using the "unified stomatal model" proposed by Medlyn et al. [33].

The microclimatic gradients within the plant canopy are explicitly accounted for. The photosynthetic active (PAR) and near-infrared (NIR) radiation, and the long-wave balance are computed for each canopy layer as in Zhao and Qualls [34,35]. The profiles of air CO2 and H2O concentrations, TA and wind speed (U) are computed using first-order turbulence closure schemes iteratively with solution of leaf energy and CO2 balance. The above-ground and soil processes are coupled through water and heat fluxes via feedbacks between soil and vegetation (rainfall interception, root water uptake, feedbacks to leaf physiological parameters).

#### *2.2. Metrics Evaluating Thermal Risk and Assimilation Capacity*

To explore the role of key plant traits and identify potential trade-offs between photosynthesis and thermoregulation, we focused on four metrics of thermal damage, tolerance and CO2 fixation.

#### 2.2.1. Maximum TL (TL,max)

TL,max was determined as the maximum leaf temperature within the warmest three-day period during the growing season. For each time step (30 min), we first calculated the average of TL of each canopy layer in sunlit leaves, which are the ones most likely to experience the warmest temperatures. Using this average TL series, we then computed the three-day moving average. TL,max is the absolute maximum TL within the three-day window with the highest moving average. We selected a three-day period, as in O'Sullivan et al. [3], considering that plants can cope with high temperatures for short periods. Knowledge of TL,max allows determining whether conditions may be conducive to thermal damage of the photosynthetic machinery, i.e., if the critical temperature (TCRIT) is exceeded over the growing season.

#### 2.2.2. Cumulated Anet (Anet,cum)

Anet,cum is the cumulated Anet within the growing season. We first computed the average of Anet at each time step (30 min), including both sunlit and shaded leaves, weighted by sunlit and shaded fractions of leaves at each layer and accounting for non-uniform leaf area distribution across canopy layers. Then, Anet,cum was calculated as the cumulated sum of the average Anet within the growing season.

#### 2.2.3. Maximum Anet (Anet,max)

The Anet,max refers to the maximum average Anet at each time step (30 min) over the growing season, weighted by sunlit and shaded fractions of leaves at each layer and accounting for non-uniform leaf area distribution across canopy layers.
