2.1 Site description and experimental design

The current study was conducted at the Oak Ridge National Laboratory's SPRUCE project site located at the U.S. Forest Service's Marcell Experimental Forest, in Minnesota, USA (47°30.476′ N; 93°27.162′ W). At the SPRUCE site, the dominant tree species is Picea mariana (black spruce) mixed with less abundant Larix laricina (tamarack). The SPRUCE experiment uses five temperature treatments (ambient or +0, which serves also as the control, +2.25°C, +4.5°C, +6.75°C, and +9°C above the ambient) established in a regression-based design (Hanson et al., 2017). These temperature treatment levels are controlled within 10 large octagonal open-top enclosures with an interior surface area of 114.8 m², and a sampling area of 66.4 m². In addition to the temperature treatments, there are also elevated CO₂ treatments, with five enclosures exposed to an ambient-CO₂ atmosphere, while the other five have an elevated CO₂ atmosphere that range between +430 and 500 ppm above the ambient. The whole-ecosystem warming treatments were initiated 15 August 2015, while the CO₂ treatments were introduced a year later, on 15 June 2016. Overall, the targets of the temperature treatment levels and CO₂ concentrations were successfully achieved (Supporting Information S1: Figure S1).

2.2 Plant material sampling and gas exchange measurements

The data used in this study were conducted between 18–30 June and 15–30 August 2017. The daytime temperatures (4:00–20:30) recorded by a climate station established at the SPRUCE site were 18.97°C and 18.02°C, for June and August, respectively. We studied individuals of 1.9–7.6 m height of the two mixed-age (up to ~45 years old) canopy tree species, black spruce and tamarack. For black spruce, one branchlet of each tree and in each plot was collected and 1-year old needle cohorts (i.e., developed in 2016) from each branchlet was measured. For tamarack, fully expanded current year foliage (i.e., developed in 2017) was used. In the June field campaign, three trees in each plot were sampled, but we decided to reduce the number of harvested branches down to two trees in the August campaign, to reduce the damage that may be caused by overharvesting in this long-term experiment. For tamarack, the same number of branchlets from different trees in each plot was used, except in one plot (in ambient CO₂ and +0) where there was only one tamarack tree to be sampled. All the data were collected on sun-exposed branchlets cut using a pruning pole. After cutting, branchlets were placed in water, and recut under water to avoid xylem transport disruption and stomatal closure.

The collection of branches took place during the early morning hours, specifically between 4 and 5 am on the day of measurement. Subsequently, these branches were carefully placed in water-filled containers within a plastic cooler. They were then transported from the field site located in Marcell, Minnesota to specialized walk-in growth chambers situated at the University of Minnesota in St. Paul, where the subsequent measurements were carried out. Before starting the measurements, branchlets were re-cut again under water. It has been demonstrated that the impact of cutting and the duration between cutting and gas exchange measurements do not exert a significant influence on stomatal conductance in conifers (Akalusi et al., 2021; Dang et al., 1997). Between 10:00 and 20:00 during the June and August field campaigns, seven portable photosynthesis systems (Li-COR 6400 XT, 6400-18 RGB light source, and 6400-22 opaque conifer chamber; LI-COR Biosciences) were employed to carry out gas exchange measurements. Measurements of net CO₂ assimilation rates (A) were conducted under saturating light conditions at 1800 µmol m⁻² s⁻¹, with variations in air CO₂ concentrations performed sequentially (400, 300, 200, 50, 400, 500, 600, 800, 1200, 1600, and 2000 µmol mol⁻¹) to generate A–C_i curves. These A–C_i curves were conducted at five different leaf temperatures (T_leaf) in the following order: 15°C, 25°C, 32.5°C, 40°C and 45°C, which produced 96A – C_i temperature response curves after data quality checks (Dusenge, Ward, et al., 2020). To attain the desired T_leaf for each target, we conducted all measurements within a walk-in chamber. This setup ensured that both the entire branch and the Li-COR IRGA (Infrared Gas Analyser) sensor were exposed to the specified temperature for a minimum of 30 min before start-up measurements at that particular temperature. In addition, this approach effectively reduced measurement errors caused by the internal thermal gradient which occurs when the Li-COR and the leaves are exposed to different temperatures, a phenomenon that had been previously documented with LI-6400 instruments (Garen et al., 2022).

Following the gas exchange measurements, we utilized ImageJ software (NH) to calculate the projected leaf area of the sampled needles. Subsequently, we adjusted for the needle area before conducting the analyses of the gas exchange data. Furthermore, we examined the collected needle tissues for elemental nitrogen concentrations (N) using equipment from Costech Analytical Technologies, Inc.

2.3 Parameterization of photosynthetic capacity

2.4 Statistical analyses

To evaluate the effects of temperature and elevated CO₂ on photosynthetic capacity (V_cmax and J_max) and their ratios at various conditions—including the reference temperature of 25°C (V_cmax25, J_max25 and J_max25/V_cmax25), the thermal optimum of each process (V_cmaxopt and J_maxopt), and the mean growth temperature (V_cmaxg and J_maxg)—as well as leaf nitrogen and leaf mass per unit area (LMA), we started with a linear mixed-effect model. In this model, warming and elevated CO₂ treatments, along with species, were designated as fixed effects, while the month during which the campaign was conducted was designated as the random effect. All analyses were run on the plot means with n = 1–4 trees/plot. The process of choosing the ultimate statistical model occurred in two stages, following the methodology outlined by Zuur et al. (2009). We first evaluated whether a random factor was required by comparing the model with the random intercept (i.e., month) and the one without any random structure. For this first step, we used the gls function in the model without the random structure and the lme function with the “Restricted maximum likelihood—REML” method in the model with the random structure, and both gls and lme functions are from the nlme R package (Pinheiro et al., 2023). We excluded the model featuring both random slope and intercept structures from the comparison, as our preliminary analyses suggested that it was over-parameterized. Thereafter, the model with the adequate random structure was selected based on the lowest akaike information criterion (AICc) (Supporting Information S1: Tables S1 and S2). When the model with random structure was not significantly different from the model without the random structure, we proceeded the analyses with a simple linear regression model. Following the determination of an appropriate random structure, we proceeded to choose the appropriate fixed effect structure. This involved considering both the model with only the main effects (i.e., warming, elevated CO₂, and species) without interactions and models incorporating all possible combinations of main effects and their interactions. The latter selection was done by comparing models with different structures using the “maximum likelihood—ML” method within the gls function from the nlme R package (Pinheiro et al., 2023). Similarly, the best fixed effect structure was selected based on the lowest AICc value using AICcmodavg R package (Mazerolle, 2023) (Supporting Information S1: Tables S1 and S2). Last, three-way repeated measures ANOVA tests were used to analyse the main effects of growth temperature, growth CO₂, leaf temperature and their interactions on the responses of V_cmax, and J_max for each species using the lmerTest package (Kuznetsova et al., 2017). The random structure of this three-way repeated measures ANOVA involved leaf temperature nested within each measured tree, which was in turn nested within each experimental plot, and further nested within each month. To derive p value and respective statistical significance for each factor and interactions, we used the Type II Wald F tests with Kenward–Roger degrees of freedom (DF). All analyses were conducted in R (R Core Team, 2024).

3.1 Temperature responses

The temperature response of V_cmax was significantly affected by warming in both species (Supporting Information S1: Figures S2, S3; Table S4). Furthermore, V_cmax was affected by warming differently when estimated at different reference temperatures. At a standard temperature of 25°C (V_cmax25), V_cmax25 was not affected by warming in both species (Figure 1a,b; Supporting Information S1: Table S3). By contrast, V_cmax estimated at both the thermal optimum (V_cmaxopt; Figure 2a,b) and at growth temperature (V_cmaxg; Figure 3a,b), significantly increased with warming in both species (Supporting Information S1: Table S3). Across species, V_cmaxopt increased by 3.3 µmol m⁻² s⁻¹ per 1°C warming (Supporting Information S1: Table S3), while V_cmaxg increased by 3.65 μmol m⁻² s⁻¹ per 1°C warming (Supporting Information S1: Table S3).

By contrast, the temperature response of J_max was largely not affected by growth temperature in both species (Supporting Information S1: Figures S4, S5; Table S4). However, similar to V_cmax, J_max also responded differently when estimated at different reference temperatures. J_max at 25°C, J_max25, was not affected by warming in both species (Figure 1c,d; Supporting Information S1: Table S3). Across the two species, J_max at the thermal optimum, J_maxopt, increased by 2.2 μmol m⁻² s⁻¹ per 1°C warming (Figure 2c,d; Supporting Information S1: Table S3), while J_max at the prevailing growth temperature, J_maxg, marginally (p = 0.064) increased by 2 μmol m⁻² s⁻¹ per 1°C warming (Figure 3c,d; Supporting Information S1: Table S3).

The ratio of J_max to V_cmax mirrored that of both V_cmax and J_max. At 25°C, warming did not affect this ratio (Figure 4; Supporting Information S1: Table S3). By contrast, the J_max to V_cmax ratio at both their thermal optima (J_maxopt/V_cmaxopt) and prevailing growth temperature (J_maxg/V_cmaxg) decreased with warming (Figure 4). Specifically, J_maxopt/V_cmaxopt decreased by 0.025 per 1°C warming, while for the J_maxg/V_cmaxg, the slope was slightly steeper, decreasing on average by 0.032 per 1°C warming across the two species (Supporting Information S1: Table S3).

3.2 Elevated CO₂ responses

The temperature response of both V_cmax and J_max was not significantly affected by elevated CO₂ in either species (Supporting Information S1: Figures S2–5; Table S4). Moreover, responses of both V_cmax and J_max estimated at any of the three reference temperatures (25°C, thermal optima and growth temperature) were consistent across the two species, as they were all significantly lower in plants grown under elevated CO₂ growth conditions compared to those grown under ambient CO₂ (Figures 1-3; Supporting Information S1: Table S3). The V_cmax25, V_cmaopt, and V_cmaxg decreased by 28%, 36%, and 35%, respectively, in trees grown under elevated CO₂ compared to their ambient-CO₂ counterparts. For J_max, both J_max25 and J_maxopt significantly decreased by 15% and 19%, respectively, while J_maxg decreased by 14%, but this decrease was not statistically significant, potentially driven by observed substantial variations in this trait (p = 0.12; Figures 1-3; Supporting Information S1: Table S3). The ratio of J_max25/V_cmax25, J_maxopt/V_cmaxopt, and J_maxg/V_cmaxg increased by 16%, 18%, and 17%, respectively, in elevated CO₂-grown trees compared to those growing under ambient CO₂ conditions (Figure 5; Supporting Information S1: Table S3).

3.3 Leaf nitrogen and leaf mass per unit area

Warming and elevated CO₂ treatments had similar effects on leaf nitrogen concentration on area basis (N_a) and leaf mass per leaf area (LMA) in both tamarack and black spruce (Figure 5; Supporting Information S1: Table S3). Leaf N_a was significantly impacted by warming and elevated CO₂ (without interaction) (Figure 5a,b; Supporting Information S1: Table S3). In response to warming, leaf N_a increased on average 0.043 g m⁻² per 1°C of warming across the two species. In response to elevated CO₂, leaf N_a was, on average, 20% lower in elevated CO₂-grown trees compared to those in ambient CO₂ conditions across the two species. By contrast, LMA was significantly affected by elevated CO₂ but not by warming across species (Figure 5c,d; Supporting Information S1: Table S3). Across the two species, LMA was, on average, 18% higher in elevated CO₂ trees than in ambient CO₂ counterparts.

3.4 Relationship between photosynthetic capacity and leaf nitrogen

Both V_cmax25 and J_max25 were positively related to leaf N_a in both species (Figure 6; Supporting Information S1: Table S4). To further explore whether leaf nitrogen have played a role in the response of photosynthetic capacity at a common leaf temperature of 25°C to the treatments, we normalised V_cmax25 and J_max25 to leaf N_a. In both species, both V_cmax25/N and J_max25/N did not significantly vary across warming treatments (Figure S6; Supporting Information S1: Table S3), suggesting that the lack of thermal acclimation of photosynthetic capacity could not largely be attributed to warming-induced effects on leaf N_a. By contrast, V_cmax25/N and J_max25/N were not significantly different between ambient and elevated CO₂-grown trees (Supporting Information S1: Table S3), suggesting that acclimation of both V_cmax25 and J_max25 to elevated strongly influenced changes in leaf N_a in elevated CO₂-grown trees.

4.1 Temperature responses of photosynthetic capacity

In both species, V_cmax25 and J_max25 did not significantly change with warming (Figure 1), partly rejecting our first hypothesis (H1), which proposed that both V_cmax25 and J_max25 should decrease with warming according to the least-optimality framework (Smith & Keenan, 2020; Wang et al., 2020). However, our results are in agreement with findings from meta-analyses (Crous et al., 2022; Kattge & Knorr, 2007; Kumarathunge et al., 2019; Way & Oren, 2010) and also those from recent experimental warming field studies on boreal and temperate seedlings (Bermudez et al., 2021; Stefanski et al., 2020), and European mature boreal conifers (Lamba et al., 2018). Nitrogen (N) is a key nutrient influencing variation of V_cmax and J_max rates (Ellsworth et al., 2022), and changes in leaf N were shown to be related to the thermal acclimation of V_cmax and J_max in some studies (Crous et al., 2018; Dusenge, Madhavji, et al., 2020, 2021; Scafaro et al., 2017; Way & Sage, 2008a). In our study, leaf N increased with warming, a response likely influenced by increases in soil nitrogen availability due to warming, as recently reported at our experimental site (Iversen et al., 2023). However, after normalizing V_cmax25 and J_max25 to leaf N (Supporting Information S1: Figure S6), photosynthetic capacity exhibited similar response to non-normalized values (i.e., no change), indicating that nitrogen have relatively little influence on thermal acclimation of these two processes. Recently, Stefanski et al. (2020) and Bermudez et al. (2021) suggested that the lack of thermal acclimation of V_cmax25 and J_max25 commonly observed in boreal experimental warming field studies may largely be due to modest warming treatments (~3 – 4°C above ambient) typically applied (Bermudez et al., 2021; Lamba et al., 2018; Smith et al., 2020; Stefanski et al., 2020). However, our study, which utilized incremental warming of 2.25°C up to 9°C—with the 9°C warming (and atmospheric CO₂ of >800 ppm) considered a likely scenario under the business as usual in some higher latitude regions by 2100 (IPCC, 2021), indicates that the lack of discernable acclimation of photosynthetic capacity to warming in realistic field conditions may be less dependent on the level of warming applied. Future studies that investigate warming effects on underlying photosynthetic biochemistry (e.g., Rubisco content and activation state, and photosynthetic pigments content) in field settings are still needed to fully understand the causes of this commonly observed lack of thermal acclimation of photosynthetic capacity at a given standard temperature, which also contradicts predictions of the least-cost optimality framework.

In contrast to photosynthetic capacity measured at 25°C, photosynthetic capacity measured at both the thermal optima of the respective process components (V_cmaxOpt and J_maxOpt) and prevailing growth temperature increased with warming (Figures 2 and 3), suggesting a positive acclimation response to warming in the studied species (Way & Yamori, 2014). These findings partly support H1, which proposed that both V_cmaxg and J_maxg should increase with warming according to the least-optimality framework (Smith et al., 2019; Smith & Keenan, 2020; Wang et al., 2020), and are consistent with results from previous studies (Scafaro et al., 2017; Smith & Dukes, 2017; Smith & Keenan, 2020; Way & Oren, 2010). This observed positive acclimation of photosynthetic capacity at the thermal optima and growth temperature suggests that V_cmax and J_max have acclimated to warming despite not being seen at the commonly used standard temperature of 25°C. In our companion SPRUCE study (Dusenge et al., 2023), we also found that the thermal optima of V_cmax and J_max in these species increased with warming, indicating a strong thermal acclimation of photosynthetic biochemistry in these boreal conifers. Overall, our findings indicate that the ability to detect the thermal acclimation of V_cmax and J_max depends on the chosen reference temperature and parameter of interest, since the shape of instantaneous temperature response of photosynthetic capacity and its acclimation response to increased growth temperature can be seen through several important temperature sensitivity parameters such as the activation energy (initial slope of the temperature response; E_aVcmax and E_aJmax), thermal optima (T_optVcmax and T_optJmax), and deactivation energy (Kumarathunge et al., 2019; Way et al., 2014; Yamori et al., 2014) of these key processes.

The ratio of J_max to V_cmax (J_max/V_cmax) is another key parameter used for modelling photosynthesis within TBMs (Rogers, Medlyn, et al., 2017). At 25°C, J_max/V_cmax was constant across warming treatments, however, at both the thermal optima of the two processes and at prevailing growth temperature, J_max/V_cmax decreased with warming (i.e., acclimated) (Figure 4). Our results at 25°C, therefore, contradict findings from several studies (Bermudez et al., 2021; Dusenge et al., 2015, Dusenge, Madhavji, et al., 2020; Smith & McNellis, & Dukes, 2020; Stefanski et al., 2020) and meta-analysis (Crous et al., 2022; Kattge & Knorr, 2007; Kumarathunge et al., 2019) that commonly report a decrease in J_max25/V_cmax25 to warming. However, a few other studies also did not observe a decrease of J_max25/V_cmax25 with warming (Crous et al., 2018; Scafaro et al., 2017). Constant J_max25/V_cmax25 in our study could largely be attributed to the observed lack of effect of warming on either V_cmax25 or J_max25. However, responses at the thermal optimum and growth temperature are consistent with findings of previous studies (Murphy & Way, 2021; Smith & Dukes, 2017) and the least-cost optimality framework (Smith & Keenan, 2020). Decrease of the J_max/V_cmax with warming is proposed to be a mechanism that reduces photorespiration that simultaneously increases with rising growth temperature (Dusenge et al., 2019; Sage & Kubien, 2007) by allocating relatively more resources to Rubisco carboxylation (i.e., V_cmax) compared to the electron transport (i.e., J_max) process (Dusenge et al., 2021; Smith & Keenan, 2020).

4.2 Elevated CO₂ responses of photosynthetic capacity

In our study, photosynthetic capacity generally acclimated to elevated CO₂, by decreasing in trees exposed to elevated CO₂ regardless of the reference temperature (Figures 1-3; Supporting Information S1: Table S3), and this response agrees with our second hypothesis (H2). Both V_cmax and J_max strongly acclimated to elevated CO₂, with V_cmax showing relatively stronger acclimation response to elevated CO₂ (−28% to –36%) compared to J_max (−14% to −19%). Observed reductions in V_cmax and J_max in our study are largely consistent with findings from meta-analyses, which are dominated by studies on temperate tree species and crops (Ainsworth & Long, 2005; Ainsworth & Rogers, 2007; Leakey et al., 2009; Smith & Keenan, 2020), but partly contrast predictions of the least-cost optimality theory which predicts rather slight positive responses of J_max to elevated CO₂ (Smith & Keenan, 2020). Consequently, the ratio of J_max to V_cmax at all three reference temperatures was higher (16%–17%; Figure 4, Supporting Information S1: Table S3) in trees exposed to elevated CO₂ compared to those growing in ambient atmospheric CO₂ conditions. The latter findings are consistent with results from a few previous studies conducted mainly on seedlings in highly controlled growth conditions (Crous et al., 2013; Dusenge, Madhavji, et al., 2020; Smith & Keenan, 2020). There has been relatively fewer studies on long-term CO₂ responses of photosynthetic capacity of mature boreal conifer trees in realistic field conditions. Before our study, we are aware of only one field study on mature trees of Norway spruce which showed that V_cmax strongly acclimated to elevated CO₂ (i.e., −23%), but the study did not investigate the responses of J_max (Lamba et al., 2018). Therefore, our current study, further generally adds results on long-term acclimation of J_max and J_max/V_cmax ratio to elevated CO₂ for the mature boreal conifers.

4.3 Responses of photosynthetic capacity to combined warming and elevated CO₂

Our study did not find any interactive effect of temperature and elevated CO₂ on V_cmax and J_max at any reference temperature (Supporting Information S1: Table S3), supporting our third hypothesis (H3). Our findings also support the predictions of the least-cost optimality framework, where a strong interaction of warming and elevated CO₂ on rates of photosynthetic capacity is not expected, and that responses of photosynthetic capacity to combined warming and elevated CO₂ should be comparable to their independent effects (see Figure 4 in Smith & Keenan, 2020). Several previous studies also did not find any interactions of warming and elevated CO₂ on either V_cmax, J_max, instead showing that V_cmax and J_max at a standard temperature acclimate to: (1) warming but not elevated CO₂ (Crous et al., 2013; Dusenge et al. 2020; Ghannoum et al., 2010; Kellomaki and Wang, 1996); (2) elevated CO₂ but not warming (Fauset et al., 2019; Ghannoum et al., 2010; Lamba et al., 2018); or (3) neither of these two factors (Murphy & Way, 2021; Wang, 1996). Clearly, there is an urgent need for more studies that focus on unraveling the biochemical mechanisms underlying the responses of V_cmax and J_max to warming and elevated CO₂. Such studies are essential for improving our understanding of photosynthetic responses and modelling under both warming and elevated CO₂ growth conditions.

In summary, our study showed that the photosynthetic capacity (V_cmax and J_max) of mature trees of North American boreal conifers responded independently to warming and elevated CO₂ when exposed to both environmental factors. Our results are consistent with several previous studies that predominantly focused on younger trees, suggesting that ontogeny minimally influences these parameters' responses to global change factors. Furthermore, our results add to growing empirical evidence that the representation of photosynthesis within TBMs under both warming and elevated CO₂ could incorporate their effects independently. However, our study was confined to two boreal conifer species. Therefore, we advocate for future research encompassing a broader spectrum of boreal species, diverse plant functional types, and other biomes to understand the widespread nature of the observed responses.