A global meta-analysis of woody plant responses to elevated CO₂: implications on biomass, growth, leaf N content, photosynthesis and water relations

Abstract

Background

Atmospheric CO₂ may double by the year 2100, thereby altering plant growth, photosynthesis, leaf nutrient contents and water relations. Specifically, atmospheric CO₂ is currently 50% higher than pre-industrial levels and is projected to rise as high as 936 μmol mol⁻¹ under worst-case scenario in 2100. The objective of the study was to investigate the effects of elevated CO₂ on woody plant growth, production, photosynthetic characteristics, leaf N and water relations.

Methods

A meta-analysis of 611 observations from 100 peer-reviewed articles published from 1985 to 2021 was conducted. We selected articles in which elevated CO₂ and ambient CO₂ range from 600–1000 and 300–400 μmol mol⁻¹, respectively. Elevated CO₂ was categorized into < 700, 700 and > 700 μmol mol⁻¹ concentrations.

Results

Total biomass increased similarly across the three elevated CO₂ concentrations, with leguminous trees (LTs) investing more biomass to shoot, whereas non-leguminous trees (NLTs) invested to root production. Leaf area index, shoot height, and light-saturated photosynthesis (A_max) were unresponsive at < 700 μmol mol⁻¹, but increased significantly at 700 and > 700 μmol mol⁻¹. However, shoot biomass and A_max acclimatized as the duration of woody plants exposure to elevated CO₂ increased. Maximum rate of photosynthetic Rubisco carboxylation (V_cmax) and apparent maximum rate of photosynthetic electron transport (J_max) were downregulated. Elevated CO₂ reduced stomatal conductance (g_s) by 32% on average and increased water use efficiency by 34, 43 and 63% for < 700, 700 and > 700 μmol mol⁻¹, respectively. Leaf N content decreased two times more in NLTs than LTs growing at elevated CO₂ than ambient CO₂.

Conclusions

Our results suggest that woody plants will benefit from elevated CO₂ through increased photosynthetic rate, productivity and improved water status, but the responses will vary by woody plant traits and length of exposure to elevated CO₂.

Introduction

Atmospheric CO₂ (atCO₂) have increased globally since the industrial revolution, owing to fossil fuel combustion and land cover changes due to increasing human population and the need for rapid economic growth (Jayawardena et al. 2021). Over the past decade, atCO₂ has been increasing at an alarming rate of 2.4 μmol mol⁻¹ year⁻¹ (Li et al. 2021) and it is currently 50% higher than pre-industrial levels (Ebi et al. 2021). It is predicted that atCO₂ may rise as high as 936 μmol mol⁻¹ by the year 2100 if greenhouse gas emissions are not mitigated (Hu et al. 2018). The increase in atCO₂ has serious impacts on plant physiology, productivity, growth, water relations (Bhargava and Mitra 2021; Zhang et al. 2021) and foliage chemistry (Du et al. 2020; Farkas et al. 2021). AtCO₂, through CO₂ fertilization, directly increases growth, canopy density and biomass by enhancing photosynthesis (Baig et al. 2015) and indirectly by reducing transpiration via partial closure of stomata (Gonsamo et al. 2021). Photosynthetic upregulation and a subsequent increase in woody plant biomass result from high carbon assimilation, owing to high investment of ribulose 1,5-bisphosphate carboxylase/oxygenase (Rubisco) to carboxylation relative to oxygenation (Wang and Wang 2021a; Raubenheimer and Ripley 2022). Indeed, many short-term studies have reported an increase in C assimilation and a subsequent increase in photosynthesis under saturating light and elevated CO₂ (eCO₂), more so for C₃ species relative to their C₄ counterparts (Zhang et al. 2021; Raubenheimer and Ripley 2022). This is driven mainly by the fact that C₃ photosynthesis does not saturate at current levels of CO₂ (Singer et al. 2020). High CO₂ uptake not only increases shoot growth and biomass, but also root depth and biomass, which further promotes soil nutrient and water uptake, indirectly enhancing photosynthesis (Thompson et al. 2017).

Increased net photosynthetic rate together with reduced transpiration as a result of reduced stomatal conductance (g_s) increase water use efficiency (WUE), thereby counter-acting moisture stress in drought-stricken ecosystems (Zhang et al. 2018; Garhum et al. 2021; Farkas et al. 2021; Mathias and Thomas 2021). At a landscape scale, increased WUE could increase soil moisture, in turn extending the length of the growing season (Li et al. 2019). Thus, CO₂ fertilization and increased WUE may drive landscape-scale vegetation transitions from open to dense woody cover stature (Gonsamo et al. 2021; Raubenheimer and Ripley 2022). However, eCO₂ reduces foliage/browse quality via a phenomenon referred to as dilution effect (Du et al. 2020). This phenomenon is the depletion of leaf N content as a result of higher accumulation of non-structural carbohydrates (NSCs) and biomass (Li et al. 2019). Because a considerable proportion of leaf N is derived from Rubisco, a reduction in Rubisco content at eCO₂ also reduces leaf N (Singer et al. 2020; Kitao et al. 2021; Wang and Wang 2021b). Dilution effect and reduced Rubisco content at eCO₂ reduce herbivore diet quality, as the decline in leaf N and increased C:N ratio reduce foliage digestibility (Du et al. 2020).

Woody plant responses to eCO₂ are very important, yet poorly understood (Bellasio et al. 2018). Understanding how these plants respond to eCO₂ may guide scientists and decision-makers in deriving climate-smart mitigation strategies (Hu et al. 2018). Amongst other disciplines, forestry plays a pivotal role in climate change mitigation by increasing C sinks through afforestation/reforestation (Lefebvre et al. 2021). For this reason, woody plant responses to eCO₂ require a special scientific assessment (Wang and Wang 2021a).

Knowledge of how trees, particularly of different functional traits, will respond to temporal changes in atCO₂ is crucial (Wang and Wang 2021b). Some short-term studies hypothesize that photosynthesis will rise linearly with atCO₂, whereas others suggest that it may saturate at a certain eCO₂ concentration (Poorter et al. 2021). Testing these two hypotheses may be difficult, particularly the latter, as different tree species exhibit differential C assimilation capacities. Inter and intraspecific variation among woody plants is due to differences in phenology, leaf types, nitrogen fixation capacity and photosynthetic pathways (Mathias and Thomas 2021; Wang and Wang 2021a, b). For instance, leguminous plants are highly likely to respond more positive due to their relationship with rhizobia which facilitates nodulation in legumes, thereby increasing C sink (Singer et al. 2020). Moreover, legumes tend to establish a symbiotic relationship with arbuscular mycorrhizal fungi which in turn increases nutrient uptake, e.g. phosphorous, thereby increasing photosynthesis and biomass (Singer et al. 2020). However, photosynthetic responses to eCO₂ depend not only on plant phylogeny, but also on duration of exposure to eCO₂ (Wang and Wang 2021a). CO₂-induced photosynthetic downregulation as a result of age-dependent changes in plant physiology has been noticed, mainly in long-term experiments (Bellasio et al. 2018; Bhargava and Mishra 2021). Increase in NSCs and reduction in leaf N, rate of photosynthetic Rubisco carboxylation (V_cmax) and photosynthetic electron transport rate (J_max) over time are implicated as the main drivers of photosynthetic downregulation (Wang and Wang 2021a). The high capacity for ribulose bisphosphate (RuBP) regeneration relative to Rubisco has been reported as a cause of reduction in photosynthetic capacity, leading to eventual downregulation (Bhargava and Mishra 2021; Singer et al. 2020). Ascertaining plant responses to eCO₂ requires thorough assessment of physiological processes inherent in photosynthesis as well as mechanisms underlying these processes across a wide range of CO₂ concentrations and plant traits on long-term basis (Poorter et al. 2021). Amongst others, discerning responses of V_cmax, J_max and g_s to eCO₂ may provide insights into processes regulating CO₂ diffusion into the leaves and its use for photosynthesis.

A meta-analysis was conducted, firstly, to assess the magnitude and direction of the effects of varying eCO₂ concentrations on woody plant growth, biomass production, photosynthetic characteristics, foliage N content, and water relations. Secondly, to assess the effect of duration of woody plant exposure to eCO₂ on plant biomass, growth, physiology, foliar quality and water relations. Thirdly, to assess how woody plant functional traits modulate responses to eCO₂. We answer the following questions: (1) how does eCO₂ affect below and above-ground productivity, morpho-physiology, foliage nutrient content, and water relations of woody plants? (2) How do woody plants with different phenology, N-fixation ability and leaf characteristics respond to eCO₂?

Materials and methods

Data collection

We compiled a database through extensive online search of peer-reviewed global studies published from 1985 to 2021 that report woody plant responses to eCO₂ (Additional file 1: Fig. S1). The literature search was conducted in Scopus, Science Direct, Google Scholar, BioOne Complete and Web of Science. To qualify for inclusion in this meta-analysis, studies had to meet the following criteria: (1) the experiment conducted paired observations at eCO₂ and aCO₂ treatments; (2) experiment was conducted exclusively on woody plants, preferably, but not limited to, trees used in forestry or grow naturally in forests, woodlands, bushlands and savannas; (3) experimental plants were exposed to eCO₂ and aCO₂ at the same time; (4) growth conditions, e.g. soil physico-chemical composition and hydro-thermal conditions were similar in experimental units (pots/plots) of eCO₂ and aCO₂ treatments and (5) experimental plants were grown as mono-species or monoculture stands, otherwise if grown as mixed stand, we considered the studies where species responses were reported separately.

We obtained a total of 1566 peer-reviewed studies, of which 100 studies with 611 observations (Additional file 1: Fig. S1), reporting on 119 woody species met the selection criteria. The response variables studied included biomass (shoot, root and total; g), shoot height (SH; cm), leaf area index (LAI; m² m⁻²), light-saturated photosynthesis (A_max; μmol CO₂ m⁻² s⁻¹), maximum rate of photosynthetic Rubisco carboxylation (V_cmax; μmol m⁻² s⁻¹), apparent maximum rate of photosynthetic electron transport (J_max; μmol m⁻² s⁻¹), leaf N on an area basis (g m⁻²), carbon:nitrogen ratio (C:N), stomatal conductance (g_s; mmol H₂O m⁻² s⁻¹), transpiration rate (Tr; mmol H₂O m⁻² s⁻¹) and water use efficiency (WUE; µmol CO₂ mmol⁻¹ H₂O). The following search keywords were used: “atmospheric CO₂” or “elevated CO₂ or “rising CO₂” or “CO₂ enrichment” in combination with (1) woody plant physiology, (2) photosynthesis, (3) below and above-ground biomass, (4) shoot growth, (5) water loss or transpiration, (6) water use efficiency, (7) stomatal conductance and (8) leaf N contents. When the response variables were reported in units different from those listed in this study, the appropriate conversions were applied.

The CO₂ treatment was considered elevated when the concentration was ≥ 600–1000 μmol mol⁻¹ and ambient when it falls within a range of 300–400 μmol mol⁻¹. The eCO₂ treatment was categorized into three discrete concentrations of < 700, 700 and > 700 μmol mol⁻¹. The 700 μmol mol⁻¹ CO₂ was used as a reference scenario based on the IPCC Special Report on Emissions Scenarios (SRES A1B) which predicted that CO₂ will rise to 700 μmol mol⁻¹ in 2100, whereas < 700 and > 700 μmol mol⁻¹ represent Representative Concentration Pathway scenarios (RCP 4.5 and 8.5), respectively (Meinshausen et al. 2011). RCP 4.5 represents a scenario where CO₂ rises below 700 μmol mol⁻¹ (approximately 650 μmol mol⁻¹; Thomson et al. 2011), whilst RCP 8.5 represents a rise above 700 μmol mol⁻¹ (approximately 936 μmol mol⁻¹) in 2100 (Hu et al. 2018). These RCPs differ in that RCP 4.5 assumes a scenario where measures are put in place to mitigate gas emissions (Thomson et al. 2011), whereas RCP 8.5 assumes a business-as-usual scenario without reductions in gas emissions (Schwalm et al. 2020).

In each study, we recorded mean ($\overline{X}$), standard deviation, sample size, reference and study duration, ambient and elevated CO₂ treatments. The $\overline{X}$s were extracted directly from tables and or through digitizing figures using Engauge digitizer V 4.1 software (http://digitizer.sourceforge.net/).

This meta-analysis comprised largely of short-term studies, in which woody plants were exposed to eCO₂ for a median time of less than a year (Additional file 1: Fig. S16 to S20). For short-term studies that conducted repeated measures, we selected $\overline{X}$ of the last sampling date because the time for plant acclimatization to CO₂ chambers was very short (< 2 weeks) in some studies. These generally include studies in which woody plant seedlings were germinated or transplanted outdoors and later transferred to CO₂ chambers. However, for longer-term studies, running over a year, we applied a more conservative approach, in which we averaged the $\overline{X}$s across repeated measures (Poorter et al. 2021). The duration (length) of tree exposure to eCO₂ was recorded for each observation to study age-related responses. Thereafter, the time of exposure to eCO₂ was categorized into the following five classes: < 0.5 year (< 6 months), 0.5–1 year, > 1–2 years, > 2–3 years and > 3 years.

In factorial experiments, we selected the CO₂ treatment where covariates were set at ambient conditions. Thus, in scenarios where drought was manipulated by reducing water supply, $\overline{X}$ for well-watered scenario was selected, assuming that watering was applied close to field capacity. Moreover, when soil fertility and light were manipulated, we selected treatments where woody plants were grown at or close to optimal rate of nutrient supply under full sunlight. If different woody plant species and or subspecies or varieties of the same species were investigated in one study, the observation for each species or variety was considered as an independent case study. Woody plant species were categorized according to N-fixation ability (leguminous and non-leguminous), leaf type (compound leaves with small leaflets, needle-like leaves, narrow leaves and broadleaves) and phenology (evergreen and deciduous). The leaf classification and description is presented in Table 1.

Table 1 The description and demonstration of leaf types of woody plants.

Meta-analysis

The meta-analysis was executed in MetaXL Microsoft (MS Excel addin) version 5.3 (Barendregt et al. 2013). The log-transformed response ratios (lnRR) between treatment (eCO₂) and control (aCO₂) were computed for each response variable in each study. Thereafter, the overall mean response ratios were calculated using mixed effects models. The positive lnRR indicates increase, negative indicates decrease and zero denotes no change. The lnRR employed in this study was as follows:

$$\ln {\text{RR}} = \ln \frac{{\overline{X}_{{{\text{eCO}}_{2} }} }}{{\overline{X}_{{{\text{aCO}}_{2} }} }},$$

(1)

where $\overline{X}$_e and $\overline{X}$_a are mean values for elevated and ambient CO₂, respectively.

Here, the lnRRs were converted to percentage response as follows:

$$\mathrm{Percentage\, change }\left(\mathrm{PC}\right)=\mathrm{ (lnRR }- 1) \times 100\mathrm{\%}.$$

(2)

To assess potential bias of the studies, we analysed Spearman’s rank-order correlations between sample sizes and InRRs, with the logic that significant (p < 0.05) correlation depicts higher bias. This emanates from work done by Wang et al. (2012) which states that, studies that report large mean differences between treatment and control are highly likely to be published compared to studies reporting marginal differences. For all response variables, no significant (p > 0.05) correlations were found between response ratios and sample sizes. Thereafter, bootstrapping of data was conducted to generate the 95% confidence intervals (CIs) using 9999 iterations.

If the 95% CI overlaps with zero, the differences between eCO₂ and aCO₂ were regarded as insignificant. The significant differences between eCO₂ concentrations (n = 3), N-fixation status (n = 2), leaf phenology (n = 2), leaf types (n = 4) and duration of exposure to eCO₂ (n = 5) were affirmed if the 95% CIs did not overlap each other. The between-study variance (I²) was calculated to examine if the significance of pooled response ratios occurred by chance or due to study heterogeneity. I² was computed as:

$$I^{2} = 100\% \times \frac{{Q_{{\text{c}}} - df}}{{Q_{{\text{c}}} }},$$

(3)

where Q_c is the Cochran’s Q heterogeneity statistic and df is the degree of freedom (Higgins et al. 2003).

If I² was large (> 50%) and the p-value associated with I² was significant (p < 0.05), removal of outlier studies was conducted to reduce I² below 25% (Patsopoulos et al. 2008). To achieve this, we plotted box plots and density plots of the response ratios and applied a remove-and-replace approach, where outlier studies were removed manually and replaced by another study to maintain adequate sample size (Additional file 1: Fig. S1). For the simplicity, the outlier studies were regarded as the studies with response ratios greater than 75thQ + 1.5IQR and lower than 25thQ − 1.5IQR. Here, Q = quartiles (25 and 75th) and IQR = interquartile range. However, the scarcity of studies for some variables constrained the remove-and-replace approach. Thus, the results for the variables represented by at most five studies should be interpreted cautiously as their large CIs may lead to a type 1 error. The regression of root vs shoot biomass, root biomass vs LAI, A_max vs leaf N, A_max vs V_cmax, A_max vs J_max, A_max vs g_s, WUE vs Tr and WUE vs g_s were conducted to study bivariate relationships. The data for each pair of variables used in each relationship were extracted from the same study.

Results

Effect of varying eCO₂ concentrations

The eCO₂ significantly increased total biomass (T_b) compared to aCO₂ and T_b responses were similar across the three eCO₂ concentrations (Fig. 1a). Shoot (S_b) and root biomass (R_b) were enhanced on average by 33 and 34%, respectively, at eCO₂, but the responses were comparable across the three eCO₂ concentrations (Fig. 1b and c). Leaf area index (LAI) and shoot height (SH) were enhanced by comparable magnitude at 700 and > 700 μmol mol⁻¹, with both LAI and SH increasing twofold more at > 700 than 700 μmol mol⁻¹ (Fig. 1d and e). The eCO₂ caused a substantial decrease in leaf N, but the responses were similar across the three eCO₂ concentrations (Fig. 1f).

Fig. 1

Percentage change (± 95% CI) in biomass (a–c), leaf area index (d), growth (e) and leaf N (f) of woody plants grown at eCO₂. The whiskers denote 95% CI and the circles denote mean percentage change (MPC) between aCO₂ and eCO₂. The numbers above the major ticks of the X-axis denote number of observations. The area fill ( ) shows the trends and the magnitude of differences between the eCO₂ concentrations. The wider the area the bigger the difference between MPCs

The A_max increased significantly at 700 and > 700 μmol mol⁻¹ by comparable magnitudes of 21 and 29%, respectively (Fig. 2a). On the other hand, there was no noticeable effect of eCO₂ of < 700 and 700 μmol mol⁻¹ on V_cmax and J_max, but rather, both parameters decreased significantly at > 700 μmol mol⁻¹ (Fig. 2b and c). The g_s decreased significantly on average by 32% at eCO₂ relative to aCO₂ (Fig. 2d). However, Tr declined only at > 700 μmol mol⁻¹ (Fig. 2e). WUE increased significantly by 35 to 63% from < 700 to > 700 μmol mol⁻¹ (Fig. 2f).

Fig. 2

Percentage change (± 95% CI) in photosynthetic characteristics (a–c) and water relations (d–f) of woody plants grown at eCO₂. The whiskers denote 95% CI and the circles denote mean percentage change (MPC) between aCO₂ and eCO₂. The numbers above the major ticks of the X-axis denote number of observations. The area fill ( ) shows the trends and the magnitude of differences between the eCO₂ concentrations. The wider the area the bigger the difference between MPCs

The A_max was significantly related to leaf N (r² = 0.30, p = 0.024), J_max (r² = 0.83, p = 0.002) and g_s (r² = 0.31, p = 0.002), but not with V_cmax (p > 0.05; Fig. 3a–d). WUE was negatively related to g_s (r² = 0.12, p = 0.046) and Tr (r² = 0.46, p = 0.001) and positively related to A_max (r² = 0.57, p < 0.001; Fig. 3e–g).

Fig. 3

The relationships between photosynthesis (A_max) and leaf N, photosynthetic Rubisco carboxylation (V_cmax) and photosynthetic electron transport rate (J_max) from a to d and water use efficiency (WUE) and transpiration (Tr), stomatal conductance (g_s) and A_max from e to g

Effect of duration of woody plant exposure to eCO₂

Total biomass decreased with increase in duration of exposure to eCO₂, with T_b increasing twofold higher when trees were exposed to eCO₂ for < 0.5 year than > 3 years (Fig. 4a). Initially, shoot biomass increased by 13% from < 0.5 to > 1–2 years, after which it declined to aCO₂ levels for trees exposed for > 3 years (Fig. 4b). Root biomass showed similar trends, both increasing by great magnitude for trees exposed to eCO₂ for < 0.5 year than when exposed for longer (Fig. 4c). Leaf N varied widely over different duration of exposure to eCO₂, but differences between eCO₂ and aCO₂ disappeared when plants were exposed for > 1 year to eCO₂ (Fig. 4d).

Fig. 4

Percentage change (± 95% CI) in biomass (a–c) and leaf N (d) of woody plants over different duration (years) of exposure to eCO₂. The whiskers denote 95% CI and the circles denote mean percentage change (MPC) between aCO₂ and eCO₂. The numbers above the major ticks of the X-axis denote number of observations. The area fill ( ) shows the trends and the magnitude of differences between the different duration of exposure to eCO₂. The wider the area the bigger the difference between MPCs. Key to duration of exposure: 0.5 years denotes half of a year (6 months). There were no observations reported for a period > 2–3 years except for leaf N

A_max varied over duration of exposure to eCO₂, with significant increase observed when trees were exposed for < 0.5 year (Fig. 5a). Generally, eCO₂ increased A_max by 30% for trees exposed for > 2–3 years, whereas it increased by 10% for trees exposed for > 3 years (Fig. 5a). V_cmax and J_max depicted similar response patterns, being low for trees exposed to eCO₂ for < 0.5 year than when exposed for longer (Fig. 5b and c). Stomatal conductance declined significantly at eCO₂, but the duration of exposure to eCO₂ had no effect on g_s (Fig. 5d). Similarly, transpiration did not differ across duration of exposure to eCO₂ (Fig. 5e). However, WUE increased significantly by 65% for the trees exposed for < 0.5 year compared to trees exposed for > 1–2 years (27%). There were no differences observed for trees exposed for < 0.5 year and other duration of exposure (Fig. 5f).

Fig. 5

Percentage change (± 95% CI) in photosynthetic characteristics (a–c), stomatal conductance (d) and water relations (e–f) of woody plants over different duration (years) of exposure to eCO₂. The whiskers denote 95% CI and the circles denote mean percentage change (MPC) between aCO₂ and eCO₂. The numbers above the major ticks of the x-axis denote number of observations. The area fill ( ) shows the trends and the magnitude of differences between duration of exposure to eCO₂. The wider the area the bigger the difference between MPCs. Key to duration of exposure: 0.5 years denotes half of a year (6 months)

Effect of woody plant traits

N-fixation ability had a significant effect on T_b, with leguminous trees exhibiting an increase of 38% compared to non-leguminous trees (27%) at eCO₂ (Fig. 6a). Although N-fixation ability did not affect S_b and R_b, leguminous trees invested more on S_b, whereas non-leguminous trees invested on R_b at eCO₂ (Fig. 6d and g). The leaf phenology had a significant effect on SH, with deciduous trees exhibiting eightfold increase in SH than evergreen trees at eCO₂ (Fig. 7b). Leaf N decreased significantly twofold in non-leguminous trees than legumes (Fig. 7g) and evergreen than deciduous trees (Fig. 7h). Needle-like leaves attained two- to fourfold higher decrease in leaf N than other leaf types (Fig. 7i).

Fig. 6

Percentage change (± 95% CI) in biomass of woody plants with different N-fixation ability (a, d and g), leaf phenology (b, e and h) and leaf types (c, f and i) grown at eCO₂. The whiskers denote 95% CI and the circles denote mean percentage change. The numbers above the major ticks of the X-axis denote number of observations. Key to leaf types: Compound-small = compound leaves with small leaflets and Needle-like = needle-like leaves

Fig. 7

Percentage change (± 95% CI) in shoot height, LAI and leaf N of woody plants with different N-fixation ability (a, d and g), leaf phenology (b, e and h) and leaf types (c, f and i) grown at eCO₂. The whiskers denote 95% CI and the circles denote mean percentage change. The numbers above the major ticks of the x-axis denote number of observations. Key to leaf types: Compound-small = compound leaves with small leaflets and Needle-like = needle-like leaves

Numerically, leguminous trees attained twofold higher increase in A_max than non-leguminous trees at eCO₂ (Fig. 8a). The A_max was not enhanced by eCO₂ for deciduous trees, whereas it increased by 21% for evergreen trees (Fig. 8b). Compound leaves with small leaflets (27%) and broad leaves (16%) attained more increase in A_max, whereas needle-like and narrow leaves were unresponsive to eCO₂ (Fig. 8c). The N-fixation ability significantly affected the responses of V_cmax to eCO₂, with non-leguminous trees exhibiting threefold decrease than leguminous trees (Fig. 8d). The V_cmax decreased significantly for evergreen trees at eCO₂, but there was no marked difference between evergreen (− 28%) and deciduous trees (− 26%; Fig. 8e). The compound leaves with small leaflets exhibited a decrease in V_cmax, whereas other leaf types were unresponsive to eCO₂ (Fig. 8f). Leguminous trees showed greater decrease in J_max and neither leaf phenology nor leaf type had a significant effect on J_max at eCO₂ (Fig. 8g).

Fig. 8

Percentage change (± 95% CI) in photosynthetic characteristics (A_max, V_cmax and J_max) of woody plants with different N-fixation ability (a, d and g), leaf phenology (b, e and h) and leaf types (c, f and i) grown at eCO₂. The whiskers denote 95% CI and the circles denote mean percentage change. The numbers above the major ticks of the X-axis denote number of observations. Key to leaf types: Compound-small = compound leaves with small leaflets and Needle-like = needle-like leaves

The g_s decreased significantly at eCO₂, more so for leguminous (− 47%) than non-leguminous trees (− 27%; Fig. 9a). The responses of deciduous and evergreen trees on g_s were comparable (Fig. 9b). The decrease in g_s at eCO₂ was significant for broad leaves and compound leaves with small leaflets (Fig. 9c). Transpiration rate was unresponsive to eCO₂ for leguminous and deciduous trees, particularly those bearing compound leaves with small leaflets (Fig. 9d–f). WUE was higher for non-leguminous (69%) than leguminous trees (46%), evergreen (69%) than deciduous trees (52%; Fig. 9g and h) and broad leaves than other leaf types (Fig. 9i).

Fig. 9

Percentage change (± 95% CI) in stomatal conductance, transpiration rate and water use efficiency of woody plants with different N-fixation ability (a, d and g), leaf phenology (b, e and h) and leaf types (c, f and i) grown at eCO₂. The whiskers denote 95% CI and the circles denote mean percentage change. The numbers above the major ticks of the X-axis denote number of observations. Key to leaf types: Compound-small = compound leaves with small leaflets and needle-like = needle-like leaves

Discussion

Woody plant responses to varying eCO₂ concentrations

This study indicated that eCO₂ enhances both shoot and root biomass production of trees growing at eCO₂ and this phenomenon has also been reported elsewhere (Ainsworth and Long 2005; de Graff et al. 2006; Wang et al. 2012). These responses are more prevalent in juvenile trees because young trees are more responsive to CO₂ and exhibit exponential growth than older trees (Pinkard et al. 2010; Wang et al. 2012). In more than 90% of studies in this meta-analysis, trees were exposed as seedlings to eCO₂, which substantiates higher biomass responses to eCO₂. Otherwise, older trees would respond differently, as their photosynthesis declines with age, since they are no longer exhibiting active vigorous growth (Walker et al. 2020). As woody plants mature, more C is invested in non-photosynthetic structures, resulting in reduced photosynthetic capacity which reduces biomass production (Curtis and Wang 1998). The root and shoot biomass were similar across the three eCO₂ concentrations, indicating that above and below-ground biomass increase regardless of the degree of rise in CO₂. The positive relationship between root biomass and shoot biomass, and LAI indicates that C allocation to roots plays a big role in increasing woody canopies (Additional file 1: Fig. S3). The higher root biomass is not only important as a C sink, but also for soil water and nutrient uptake (Thompson et al. 2017; Wang and Wang 2021b). Plants that produce more roots, largely deep-rooted trees have an advantage to access ground water during periods of moisture stress and drought (Uddin et al. 2018). While higher biomass investment on leaves increases C sequestration, the residence time of C might be shorter relative to roots (Walker et al. 2020). Leaf area index was similar between 700 and > 700 μmol mol⁻¹ (Fig. 1d), but shoot height was taller at > 700 μmol mol⁻¹ (Fig. 1e), suggesting that CO₂ might increase stem elongation without significant effects on tree canopy sizes if eCO₂ increases above 700 μmol mol⁻¹. The lack of increase in LAI at > 700 μmol mol⁻¹ was likely due to the lack of increase in the rate of photosynthesis as depicted by similar A_max at 700 and > 700 μmol mol⁻¹ (Fig. 2a).

The leaf N content decreased markedly across eCO₂ concentrations, which may be ascribed to N dilution by accumulation of secondary compounds, as depicted by increase in the C:N ratio at eCO₂ (Additional file 1: Fig. S4). Our results further indicated that, regardless of the extent of increase in CO₂ in the future, leaf N will decrease by almost the same magnitude. The eCO₂-driven decrease in leaf N (18–25%) in this study is higher than 16 and 12% decrease reported by Curtis and Wang (1998) and Jayawardena et al. (2021), respectively. In this study, the percentage change in leaf N was calculated from mean N content of the last sampling date for short-term studies, of which for most studies, this time was towards the end of the growing season. Thus, the decline in leaf N could be ascribed to senescing leaves, lack of replacement of older leaves by new ones and N translocation to below-ground plant parts (Tom-Dery et al. 2019).

A decline in leaf N implies that herbivores will depend largely on N-deficient foliage (Coley et al. 2002) and this may need a change of feeding habits (Farkas et al. 2021) and increased foliage intake to compensate for N deficiency (Jayawardena et al. 2021). A decline in leaf N content implied that eCO₂ may reduce decomposition rate of the leaf litter as well as N cycling (Norby et al. 1999). These effects have serious implications not only for herbivores, but also for plant nutrition because N deficiency in the soil may hinder plant growth. The increase in leaf C:N ratio is consistent with Du et al. (2020) in their recent meta-analysis of the responses of leaf nutrients to eCO₂.

The V_cmax and J_max declined at eCO₂ indicating that rise in CO₂ downregulates carboxylation rate of Rubisco and electron transport rate. A depletion in maximum carboxylation rate of Rubisco at eCO₂ has been reported widely by previous meta-analytic studies working not only with trees, but also with crops and grasses (Wang et al. 2012). However, A_max was enhanced by a similar degree at 700 and > 700 μmol mol⁻¹, indicating that downregulation of V_cmax and J_max did not completely negate photosynthesis. Positive relationships between A_max and J_max, leaf N and g_s (Fig. 3) indicated that photosynthesis was, in fact, controlled by biochemical processes and stomatal aperture at eCO₂. The positive relationship between A_max and g_s indicates that despite decline in g_s at eCO₂ concentrations, this did not lead to stomatal limitation of photosynthesis. A reduction in conductance may limit C assimilation, more so in needle-like leaves which have dense leaves and thick cell walls of the photosynthetic cells that may limit diffusion of CO₂ into the chloroplast (Guo et al. 2022). A positive relationship between g_s and A_max was also reported by Medrano et al. (2002) and Guo et al. (2022). Since CO₂ activates Rubisco, increased CO₂ diffusion into the active site of Rubisco modulated by g_s means increase in V_cmax and thus enhanced A_max. For example, when V_cmax was similar between eCO₂ and aCO₂, A_max increased at 700 μmol mol⁻¹, but when V_cmax decreased at > 700 μmol mol⁻¹, A_max did not increase anymore (Fig. 2a and b). The relationship between V_cmax and J_max (Additional file 1: Fig. S2), as has been reported in other studies, e.g. Gardner et al. (2021) and Byeon et al. (2021), indicated that enhancement of A_max at eCO₂ was a function of a coordination between V_cmax and J_max (Yang et al. 2021). On the other hand, a coupling between leaf N and A_max was not surprising (Fig. 3a), given that 15–35% of leaf N is allocated to Rubisco, a key enzyme facilitating photosynthesis (Evans et al. 1989; Luo et al. 2021). In their recent model of N allocation, Bachofen et al. (2022) showed that more leaf N was partitioned to V_cmax in the uppermost and to J_max in the bottom of the tree canopy. The relationship between leaf N and A_max observed in this study suggests that more leaf N was invested in photosynthetic apparatus. Otherwise, a decoupling between leaf N and A_max would suggest a reallocation of N to non-photosynthetic machinery, which would lead to acclimation of A_max.

In our meta-analysis, the similarity in A_max between 700 and > 700 μmol mol⁻¹ suggests that eCO₂ may stimulate photosynthesis up to 700 μmol mol⁻¹, above which the rate of stimulation declines in woody plants. This leads to a speculation that, given the linear increase in atmospheric CO₂ with time, photosynthesis may acclimatize to CO₂ above 700 μmol mol⁻¹. This is supported by the trends of LAI which increased by similar magnitude at 700 and > 700 μmol mol⁻¹ (Fig. 1d), indicating limited enhancement of leaf production and photosynthesis at eCO₂ above 700 μmol mol⁻¹. This could be ascribed to saturation of Rubisco which normally occurs at eCO₂ of 700–1000 μmol mol⁻¹ at which photosynthesis is limited by ribulose‐1,5‐bisphosphate (RubP) regeneration (Bond and Midgley 2000). Similarly, Runkle (2015) attest that the effect of eCO₂ is negligible at 800–1000 μmol mol⁻¹. However, in their meta-analysis, Poorter et al. (2021) found that photosynthesis was saturated at eCO₂ above 1000 μmol mol⁻¹. Albeit they did not study tree responses, Zheng et al. (2018) found that eCO₂ above 600 μmol mol⁻¹ downregulated photosynthesis on plants grown at a wide range of 600–1600 μmol mol⁻¹. Photosynthetic downregulation was ascribed to a concurrent downregulation of V_cmax and J_max at eCO₂ (Zheng et al. 2018). However, it appears that 700 μmol mol⁻¹ as an estimate for future eCO₂ may cause uncertainty for future projections. The uncertainty of 700 μmol mol⁻¹ is expected, given unpredictable variation in CO₂ emissions in space and time (Prentice et al. 2001). Variability in future CO₂ emissions caused largely by variability in human population increase and energy demand may lead to deviation of future eCO₂ from 700 μmol mol⁻¹, thereby reducing reliability of this concentration for future projections (Prentice et al. 2001).

Stomatal conductance and transpiration rate were reduced and the WUE was increased by eCO₂. The decline in transpiration at eCO₂ was not surprising, as the eCO₂ reduces stomatal conductance and density (Kerstiens et al. 1995). The decline in g_s and Tr coupled with increase in A_max under eCO₂ as observed here facilitated increase in water use efficiency. A reduction in g_s at eCO₂ is reported in many studies (Xu et al. 2016; Baligar et al. 2021; Wang and Wang 2021a, b; Zhang et al. 2021) and this response is reported to be more advantageous in water limited areas. In their meta-analyses, Wang and Wang (2021a) and Walker et al. (2020) reported a decrease in g_s in trees grown at eCO₂. Although the decrease of g_s in the current meta-analysis is lower than that reported by Wang and Wang (2021a) and Walker et al. (2020), all these studies concur on that eCO₂ reduces g_s. The negative relationships between WUE and g_s, and Tr suggest that reduction in g_s and Tr improve water status of woody plants grown at eCO₂. The slope was steeper for transpiration (Fig. 3e), signifying that a reduction in water loss plays a more important role in increasing WUE. This may help delay the onset and reduce the degree of moisture stress (Wang et al. 2012), in turn extending the length of the growing season in forest ecosystems (Souza et al. 2019).

Woody plant responses over different duration of exposure to eCO₂

The apparent increases in woody plant biomass for trees exposed for < 0.5 year agrees with other previous studies. Woody plants, more so young plants are highly sensitive and responsive to CO₂ fertilization (Raubenheimer and Ripley 2022). This was also confirmed by greater increases in photosynthesis of trees exposed for < 0.5 year to eCO₂ (Fig. 5a), of which most were exposed as seedlings (Additional file 1). Moreover, water use and N use efficiency are high during early stages of exposure to eCO₂. These together with CO₂ fertilization promote stem elongation and girth size, leaf production and branching, thereby increasing shoot biomass (Bhargava and Mishra 2021). In this study, root biomass as well as total biomass declined with increase in duration of exposure to eCO₂ (Fig. 4a and c), indicating that responses to eCO₂ are age-dependent, as plants acclimatized to eCO₂ as they mature. Similarly, Idso (1999) showed a gradual decline in biomass of Quercus and Pinus species over a duration of 35 years, with declines commencing as early as less than 5 years of exposure to eCO₂. Our results suggested that over a long-term exposure to eCO₂, trees may exhibit sink limitations due to age-related ecophysiological changes. For root biomass, declining trends were expected given that our dataset was derived largely from pot experiments in which rooting depth probably was limited by the pot size (Curtis and Wang 1998). However, shoot biomass increased up to > 2–3 years of exposure to CO₂ (Fig. 4b), signifying that shoot biomass does not rely only on its relationship with root biomass. This result agrees with Zhang et al. (2011) who found higher shoot biomass 3 years after tree exposure to eCO₂.

A_max was enhanced during short-term exposure (< 0.5 year), beyond which A_max stimulation became insignificant, with response trend declining towards the levels of aCO₂ for trees exposed for more than 3 years. This result signifies that photosynthetic stimulation by eCO₂ is transient and that it is strongest on actively growing plants (Dusenge et al. 2019). This was confirmed by weakened effects of eCO₂, characterized by photosynthetic acclimation for woody plants exposed for > 3 years (Fig. 5a), 4 to 5 years specifically (Additional file 1: Fig. S10). However, in this study, trees exposed for > 3 years to eCO₂ were already 3–8 years old during application of CO₂ treatment. Thus, we postulate that the age at exposure to eCO₂ could have played a role in photosynthetic acclimation. Photosynthetic acclimation was noticed as early as less than a year of exposure to eCO₂ (Ainsworth et al. 2002; Hymus et al. 2002), after three growing seasons in Picea sitchensis (Centrito and Jarvis 1999) and after 10 years for Liquidambar styraciflua elsewhere (Warren et al. (2015). Eamus and Jarvis (2004) showed that photosynthesis may acclimatize as early as few months of exposure to eCO₂. Generally, as plants grow, there is more accumulation of non-structural carbohydrates (NSC) accompanied by a depletion of leaf N, which therefore reduces photosynthetic capacity at eCO₂. In this study, we show that leaf N was consistently reduced by eCO₂ over different durations of exposure, with A_max appearing to follow the leaf N trends. The downregulation of A_max as duration of exposure to eCO₂ increases is more common in evergreen species, as N in previous year’s leaves tends to be depleted relative to current year’s leaves (Medlyn et al. 1999). V_cmax and J_max were tightly coupled, depicting similar trends over duration of exposure to eCO₂. However, the coupling of these photosynthetic traits does not appear to have influenced A_max responses, as there were no obvious relationships between A_max and these parameters over different duration of exposure.

The g_s remained low and similar across different durations of exposure to eCO₂. We found no stomatal acclimation to eCO₂ over time. As a result, transpiration was reduced by almost similar magnitudes across different durations of exposure to eCO₂. Likewise, Wang and Wang (2021a, b) found no stomatal acclimation to eCO₂ in their recent meta-analysis. However, despite the lack of variation in g_s and Tr over time, WUE was highest for woody plants exposed for < 0.5 year to eCO₂, indicating that greater photosynthesis observed in trees exposed for < 0.5 year increased WUE. The numerically low WUE for trees exposed for longer (> 3 years) could be ascribed to acclimation of A_max, which probably reduced sink strength of trees, resulting in decline in shoot photosynthetic responses. Another possibly explanation for low WUE could be reallocation of photosynthates to non-photosynthetic organs, e.g. stem and roots instead of leaves, resulting in reduced photosynthesis. This is supported by abrupt decline in shoot production over 3 years of exposure to eCO₂, whereas root biomass was consistently 20% higher in eCO₂ relative to aCO₂. However, this meta-analysis was derived from short-term studies (< 10 years), hence it is unclear how woody plants exposed more than 10 years would respond to eCO₂. This may cause uncertainty in future projections of woody plant responses to future eCO₂ because plant responses to eCO₂ become weak as plants age (Wang and Wang 2021a). Thus, future research on long-term exposure of woody plants using free-air CO₂ enrichment (FACE) is warranted to unpack age-related physiological responses and provide insights into sink strength of forests over a long-term to derive precise projections for future.

Influence of woody plant functional traits on responses to eCO₂

Consistent with previous studies, the current study indicates that leguminous trees attained more biomass than non-leguminous trees at eCO₂. This was expected, given that leguminous trees use symbiosis to fix N, which gives them advantage over non-leguminous trees, especially in N-limited soils (Chen and Markhan 2021; Kou-Giesbrecht et al. 2021). Similar findings where a leguminous plant attained more biomass than non-leguminous plant grown at eCO₂ were reported by Lee et al. (2003). Moreover, Singer et al. (2020) reported that leguminous species have a close relationship with arbuscular mycorrhizal fungi which promotes efficient uptake of P, in turn increasing biomass of leguminous trees. In addition, increase in nodule mass in legumes which is non-existent in non-leguminous trees could have contributed to the total biomass production of legumes. Interestingly, more biomass was allocated to shoot in leguminous trees, whereas non-leguminous trees invested more on root production. High shoot biomass of leguminous trees could be attributable to high leaf production, as depicted by greater increases in LAI at eCO₂ (Fig. 7d), probably due to a synergy between C fertilization and N-fixation. Our findings agree with Zhang et al. (2011) who reported a substantial increase in shoot biomass and no change in root biomass of a leguminous shrub (Caragana microphylla) growing at eCO₂. High root production by non-leguminous trees may facilitate greater C input into the soil via rhizodeposition, more so if root respiration is minimal (de Graaf et al. 2006).

Increased shoot height of deciduous trees at eCO₂ than evergreen trees is not surprising because according to Poorter and Navas (2003), evergreen trees grow more slowly than deciduous trees. Deciduous species capitalize on higher photosynthesis per unit leaf mass (Zhang et al. 2021) and leaf N content which facilitate rapid growth relative to evergreen trees (Givnish 2002). Moreover, the cost of utilizing C for production of secondary compounds at the expense of photosynthesis, common in evergreen trees, suppresses shoot growth (Givnish 2002).

However, it is worth noting that this meta-analysis was derived largely from short-term experiments, hence, we report more on seedlings and or juveniles of different phenology rather than older plants. Thus, for an example, if deciduous and evergreen seedlings were grown at eCO₂ at the beginning of a growing season and shoot measurements taken at the end of the season, deciduous seedlings may gain an advantage due to: (1) greater and rapid photosynthesis they attain before dormancy and (2) evergreen seedlings might be deprived an advantage to grow during winter when deciduous seedlings are dormant. Moreover, the scarcity of studies reporting long-term growth of trees in eCO₂ limits understanding of how leaf N in older plants respond to eCO₂ relative to juveniles. Thus, the present results are inconclusive and cannot be generalized for trees exposed to eCO₂ longer than one growing season, due to age and size-dependent changes in plant physiology (Saxe et al. 1998; Garner et al. 2021).

The leaf N was higher in leguminous than non-leguminous trees, as a result, C:N was higher for the latter than the former at eCO₂ (Additional file 1: Fig. S6). Similar findings were reported in other previous studies (e.g. Du et al. 2020) and are often ascribed to the dilution effect, as a result of high accumulation of biomass and non-structural carbohydrates (Xia et al. 2021). The higher C:N in non-leguminous trees relative to leguminous trees could be explained largely by low decline in leaf N for the latter compared to the former.

Despite the lack of statistical differences on A_max responses to eCO₂ between leguminous and non-leguminous trees, the twofold increase in A_max attained by the former compared to the latter need to be considered. The A_max increased regardless of higher decrease in V_cmax and J_max for leguminous trees relative to non-leguminous trees. Similar results were reported in the meta-analysis by Wang and Wang (2021a), where V_cmax and J_max declined and A_max increased under eCO₂. High allocation of biomass to shoot could have enhanced photosynthesis in leguminous trees than in non-leguminous trees which allocate more biomass to roots (Fig. 6d and g).

The positive responses of A_max to eCO₂ were substantial for compound leaves with small leaflets, which are mostly legumes, e.g. Acacias and Prosopis species. This finding further supports higher A_max attained by leguminous trees, which is attributable to high leaf N and higher investment on shoot biomass than in non-leguminous trees. A_max was also increased in broadleaves (Fig. 8c), owing to thicker palisade mesophyll layers that keep high CO₂ in the leaves, such that even when stomata closes, photosynthesis remains high (Zhang et al. 2021).

Stomatal conductance declined for both leguminous and non-leguminous trees. However, transpiration rate remained similar between eCO₂ and aCO₂ regardless of the decline in g_s for leguminous trees, which consequently translated to low WUE compared to non-leguminous trees (Fig. 9a, d and g). This response is surprising given that stomatal closure limits water loss (Zhang et al. 2021; Li et al. 2021). It is, however, worthy to postulate that higher reduction in g_s exhibited by leguminous trees increased leaf temperature, which probably amplified diffusion out of water in the leaves (Kerstiens et al. 1995). The higher WUE for non-leguminous trees was driven mainly by a decline in Tr (Fig. 9d). As has been reported in other previous studies (e.g. Soh et al. 2019), evergreen trees had higher WUE than deciduous trees. Similarly, Zhang et al. (2021) reported higher WUE for evergreen broadleaved trees than deciduous broadleaved trees elsewhere.

In this study, higher WUE for evergreen trees could be explained by the balance between A_max and Tr, whereas deciduous trees exhibited a numerically low A_max and higher Tr. The higher photosynthesis normally shown by evergreen trees plays an important role in increasing WUE (Soh et al. 2019). The broadleaved trees attained higher WUE than other leaf types due to reduced g_s and Tr (Fig. 9), suggesting that eCO₂ in broadleaved forests may enhance soil moisture. This could be an important adaptation strategy to the future climate characterized by extreme temperatures and drought. The low enhancement of WUE for compound leaves with small leaflets is attributable to the lack of decline in Tr. On the other hand, the lack of decline in g_s for the needle-like leaves observed in this study was also reported for conifers by Saxe et al. (1998) in their systematic review. Generally, the guard cells of conifers are less sensitive to eCO₂ (Ainsworth and Rogers 2007). As a result, in this study, WUE in needle-like leaves was low, owing to high Tr caused by unresponsive behaviour of g_s to eCO₂.

Conclusions

Overall, this meta-analysis revealed that eCO₂ increases woody plant growth, productivity, photosynthetic rate and water status, but reduces foliage quality via reduced leaf N. It appeared however, that photosynthesis is enhanced to a certain degree, after which the rate of stimulation declines at eCO₂ above 700 μmol mol⁻¹, signifying that photosynthetic acclimation is likely at relatively high CO₂. This response appears to be age-dependent, as the photosynthetic acclimation was more apparent in plants exposed for a longer duration to eCO₂ than those exposed for less than a year. The increase in photosynthesis especially during early exposure to eCO₂ together with reduction in water loss were central in improving water use efficiency of woody plants. Our results further indicated that responses to eCO₂ are dependent on woody plant traits. The high biomass production and low decline in leaf N in leguminous trees at eCO₂ indicated that these woody plants may be more important as a source of forage for herbivores than non-leguminous trees. Broad leaves showed a substantial increase in water use efficiency than other leaf types, underpinning that through this water saving strategy broadleaved forests would be less vulnerable to the future extreme climate.