A meta-analysis of mesophyll conductance to CO2 in relation to major abiotic stresses in poplar species

Abstract Mesophyll conductance (gm) determines the diffusion of CO2 from the substomatal cavities to the site of carboxylation in the chloroplasts and represents a critical component of the diffusive limitation of photosynthesis. In this study, we evaluated the average effect sizes of different environmental constraints on gm in Populus spp., a forest tree model. We collected raw data of 815 A–Ci response curves from 26 datasets to estimate gm, using a single curve-fitting method to alleviate method-related bias. We performed a meta-analysis to assess the effects of different abiotic stresses on gm. We found a significant increase in gm from the bottom to the top of the canopy that was concomitant with the increase of maximum rate of carboxylation and light-saturated photosynthetic rate (Amax). gm was positively associated with increases in soil moisture and nutrient availability, but was insensitive to increasing soil copper concentration and did not vary with atmospheric CO2 concentration. Our results showed that gm was strongly related to Amax and to a lesser extent to stomatal conductance (gs). Moreover, a negative exponential relationship was obtained between gm and specific leaf area, which may be used to scale-up gm within the canopy.


Highlight 30
Our meta-analysis on the variation of mesophyll conductance under abiotic stresses 31 shows a noticeable response to light gradient, soil moisture and nitrogen availability and 32 a significant relationship with specific leaf area 33 34 Introduction 53 Carbon assimilation of plants is importantly determined by the diffusion efficiency of 54 CO 2 from the atmosphere to the site of carboxylation. The rate of CO 2 diffusion is 55 affected by two main diffusion limitations. The first limitation controls the CO 2 flux from 56 the atmosphere to the sub-stomatal cavities through the stomata and is characterized 57 by stomatal conductance (g s ). The second limitation determines the diffusion of CO 2 58 from the substomatal cavities to the sites of carboxylation in the chloroplasts and is 59 characterized by mesophyll conductance (g m ). g m is composed of gaseous and liquid 60 phase resistances (Flexas et al., 2008;Evans et al., 2009;Niinemets et al., 2009). CO 2 61 diffusion inside the leaves is complex, facing a series of structural barriers coupled with 62 biochemical regulations. It has been shown that g m is typically limited by liquid phase 63 conductance both in species with soft mesophytic leaves as well as in species with 64 tough xerophytic leaves (Tomás et al., 2013;Tosens et al., 2012a,b). The liquid phase 65 is a multi-components pathway that involves the mesophyll cell wall thickness and 66 porosity, the plasmalemma, the chloroplast envelope, the chloroplast thickness and the 67 mesophyll surface area exposed to intercellular air spaces per unit of leaf area (Evans 68 et al., 2009;Tosens et al., 2012b;Tomás et al., 2013). After extensive study during the 69 last two decades, g m is now widely accepted as a critical limiting factor to 70 photosynthesis, which has to be considered in characterizing plant carbon gain 71 potentials and responses to future climate change Niinemets et al., 72 2009; Niinemets et al., 2011;Flexas et al., 2016). 73 Mesophyll conductance has been shown to respond to environmental stress and may 74 govern functional plasticity of photosynthesis and plant fitness under limited resources 75 (Galle et al., 2009;Barbour et al., 2010;Buckley and Warren, 2014;Théroux-Rancourt 76 et al., 2015;Flexas et al., 2016;Shrestha et al., 2018). However, recent findings on the 77 response of g m to abiotic stress are conflicting and inconclusive, demonstrating the 78 complex nature of g m variation (Flexas et al., 2008;Niinemets et al., 2009;Zhou et al., 79 2014; Shrestha et al., 2018). This suggests that the environmental and species-specific 80 response (acclimation) of g m to growth conditions should be considered to predict plant 81 performance in the field. Among the contrasting environmental responses, growth 82 temperature may (Warren, 2008;Silim et al., 2010) or may not (Dillaway and Kruger,83 2010; Benomar et al., 2018) affect g m . Similarly, the increase in soil nitrogen may 84 (Warren 2004;Shrestha et al., 2018;Xu et al., 2020;Zhu et al., 2020) or may not (Bown 85 et al., 2009) stimulate g m . The magnitude of decrease in g m under water stress and low 86 light differed among studies (Warren et al., 2003;Niinemets et al., 2006;Montpied et al., 87 2009; Bögelein et al., 2012;Tosens et al., 2012a;Zhou et al., 2014;Peguero-Pina et al., 88 2015;Théroux Rancourt et al., 2015). These discrepancies among studies result in part 89 from (i) the absolute changes in anatomical, morphological (mesophyll structure) and 90 biochemical (aquaporins and carbonic anhydrase) traits controlling g m , as well as from 91 changes in the relative contribution of these traits (Marchi et al., 2008;Tomás et al., 92 2013), and from (ii) the level of coordination between g m , g s and leaf-specific hydraulic 93 conductivity (K L ) (Flexas et al., 2013;Théroux-Rancourt et al., 2014;Xiong et al., 2017). 94 Given the complex interplay between different factors controlling g m , it is important to 95 examine its acclimation at the genus and species level to gain a general insight into the 96 mechanistic basis of changes in g m . 97 Five methods exist to estimate g m : a) chlorophyll fluorescence coupled to gas exchange 98 (Harley et al., 1992), (b) carbon isotope discrimination coupled to gas exchange (initially 99 developed by Evans et al., 1986), c) oxygen isotope discrimination (Barbour et al.,100 2016), (d) A-C i curve fitting (Ethier and Livingston, 2004;Sharkey et al., 2007), and (e)

101
1D modeling of g m from leaf structural characteristics , Tosens et al., 102 2012b, Tomas et al., 2013. All of these methods are based on assumptions and each 103 one has its limitations (Flexas et al., 2013;Tosens and Laanisto, 2018). The standard 104 deviation of the estimate of g m may vary from 10% to 40%, which may limit our 105 understanding of g m acclimation to growth conditions, particularly when the variation 106 between treatments or studies is less than the error of estimates (Sun et al., 2014a).

107
Populus spp., model crops in forestry characterized by high yield potential, have been 108 the subject of numerous studies to understand the physiological response to 109 environmental factors but research is still necessary to make assessment of effects 110 sizes and to draw generalizations (Larocque et al., 2013). A general understanding of 111 the CO 2 pathway through mesophyll and how it is affected by environmental factors 112 would be beneficial in the efforts to (i) accurately predict canopy photosynthesis under 113 different environmental conditions, particularly under warmer and drier climate, and 114 improve global carbon assimilation models and (ii) develop and effectively select more 115 resilient and productive cultivars for wood, bioenergy and bioremediation. Substantial 116 data of A-C i response curves in the literature has been used to estimate photosynthetic 117 parameters, not to estimate g m , and such compiled dataset would provide a basis to 118 make such assessments on the response of g m to the environment.

119
In this study, we compiled 815 A-C i response curves from 26 datasets of different poplar 120 species and hybrids (Table 1). Published A-C i curve-fitting approaches differ broadly 121 regarding the rectangularity of the hyperbola, segmentations of the model of 122 photosynthesis and determination of the transition value of CO 2 from carboxylation to 123 electron transport (Harley et al., 1992;Ethier and Livingston, 2004;Manter and Kerrigan 124 2004;Dubois et al., 2007;Sharkey et al., 2007;Pons et al., 2009;Gu et al., 2010).

126
Although A-C i curve fitting is unreliable for species with large g m , it can provide results 127 similar to those obtained from direct measurements for species with medium to low g m 128 (Niinemets et al., 2005(Niinemets et al., , 2006Qiu et al., 2017;Xu et al., 2020). Using the compiled A-C i 129 response curves, we performed curve fitting using a single method (Ethier and 130 Livingston, 2004) to alleviate the fitting method bias and to obtain uniformed estimates 131 of g m , maximum rate of carboxylation (V cmax ) and rate of electron transport (J). We 132 further collected related variables like leaf nitrogen content, stomatal conductance and 133 specific leaf area (SLA) when data were available. Our main goal was to find trends in 134 the response of mesophyll conductance to prevalent abiotic stressors and to examine 135 the relationship between g m and other leaf physiological and morphological traits. We 136 believe that a meta-analytical approach to analyze the accumulated data on the 137 diffusion of CO 2 through the mesophyll diffusion pathway in relation to other 138 photosynthesis-related traits provides key insight into physiological and structural 139 controls on mesophyll conductance and into the environmental plasticity of mesophyll 140 conductance. We aim at contributing to the efforts of improving poplar photosynthetic 141 efficiency in poplar breeding programs, and at improving modelling of global carbon 142 assimilation of biomass and bioenergy crops under climate change.

144
Data collection 145 Data were collected by a web search in Web of Science, Scopus, and Google Scholar 146 using the following key words: ("Populus" or "poplar" or "hybrid poplar" or "aspen") and 147 ("V cmax " or "maximum rate of electron transport (J max )" or "mesophyll conductance"). At 148 this step, abstract of every item was checked to confirm the paper is actually about g m .

149
Then, we looked at the materials and methods section of selected papers where A-C i 150 response curves of Populus spp. were measured.

151
To get raw data of A-C i response curves, we contacted the corresponding authors or 152 co-authors of the targeted studies by e-mail and via ResearchGate. We obtained 23 153 data sets from published studies and three data sets from unpublished studies ( Table   154 1). Collectively, they provided a total of 815 A-C i response curves.

155
The total data of 72 genotypes were collected from measurements on plants growing in 156 plantations (5 studies), or under controlled conditions (greenhouse or growth chamber using at least two methods simultaneously never exceeded 1 mol m -2 s -1 (Singsaas et 164 al., 2004;Flexas et al., 2008;Velikova et al., 2011;Tosens et al., 2012a;Théroux-165 Rancourt et al., 2014;Momayyezi and Guy, 2017;Xu et al., 2020). Then, g m > 1 mol m -2 166 s -1 were considered as non-available data (94 entries), and V cmax and J values retained 167 for further analyses. Furthermore, fewer data point and/or large interval in the 168 intracellular concentration of CO 2 (C i µmol mol -1 ) in the RuBisCO limited phase region 169 are the major source of error and probability of failing g m (reached the set bound during 170 the grid search) frequently encountered during A-C i curve fitting (Warren, 2006;Miao et 171 al., 2009, Sun et al., 2014aMoualeu-Ngangue et al., 2016).

173
To examine the effect of a given abiotic factor on g m , we estimated that a minimum of 174 three studies is necessary to have reliable conclusions, regardless of the genotype 175 used. Then, we could come up with subsets of data that focused on the same variable 176 and performed analyses on them separately (identified in the column 'treatment' in 177   Table 1). Our first goal was to examine the effect of variations in these factors on g m,

178
light-saturated photosynthetic rate (A max ), g s , J, V cmax and in a second step, the 179 relationships between g m and other photosynthetic characteristics (A max , g s , J, V cmax ).

189
-Copper stress: data sets from studies of Borghi et al. (2007) and Borghi et al. 190 (2008) were used to examine the response of poplar trees to contamination of 191 the substrate with copper (Cu). Treatments were assigned to three levels of Cu:    (Table 1).

207
For studies with two or more investigated factors, we considered the different levels of 208 the factor of interest and the control level of the rest of factors to avoid between-factors 209 interaction effects on results. For example, in Calfapietra et al. (2005), trees were 210 subject to different levels of N and CO 2 . When we focused on the effect of N, we 211 selected trees exposed to ambient CO 2 only (control).

212
Curve analysis 213 Mesophyll conductance and photosynthetic capacity variables, V cmax and J, were 214 estimated by fitting A-C i curve with the non-rectangular hyperbola version of the 215 biochemical model of C 3 plants (Farquhar et al., 1980). The model was fitted using non-

218
Briefly, the net assimilation rate (A n ) is given as: where V cmax is the maximum rate of carboxylation (µmol m -2 s -1 ), O is the partial 225 atmospheric pressure of O 2 (mmol mol -1 ) , Γ * is the CO 2 compensation point in the 226 absence of mitochondrial respiration, R day is mitochondrial respiration in the light (µmol 227 CO 2 m -2 s -1 ), C c is the chloroplast CO 2 (µmol mol -1 ), C i is the intercellular air space 228 concentration of CO 2 (µmol mol -1 ), K c (µmol mol -1 ) and K o (mmol mol -1 ) are the 229 Michaelis-Menten constants of RuBisCO for CO 2 and O 2, respectively, J is the rate of 230 electron transport (µmol m -2 s -1 ). The values at 25 °C used for K c , K o and Γ * were 272 231 µmol mol -1 , 166 mmol mol -1 and 37.4 µmol mol -1 , respectively (Sharkey et al., 2007) and 232 their temperature dependencies were as in Sharkey et al. (2007).

233
In four data sets, measurements were carried out under a temperature that was 234 different from the reference (25 °C). In this case, V cmax and J were normalized to 25 °C 235 using the model of Kattge and Knorr (2007), which integrates the acclimation to growth 236 temperature. However, the actual values of V cmax and J were more often significant 237 compared to normalized values, and this was true using both ANOVA and regression.

242
When at least three studies focused on one factor (nitrogen, CO 2 , canopy level, 243 copper), the effect of treatments on light-saturated photosynthetic rate (A max ), g m , and g s 244 was assessed, separately for each response variable, through mixed model analyses of 245 variance using the primary data (Riley et al., 2010;Mengersen et al., 2013).

246
"Treatment" was the fixed effect while "study" and "genotype" nested within study were 247 the random effects. The number of replicates was not necessarily balanced across 248 treatments. The assumptions of normality of the residuals and homogeneity of variance 249 were verified, and a log-transformation was made when necessary.

251
The number of studies on mesophyll conductance has rapidly increased since 2000, Increased air CO 2 had no effect on average A max (14.43 ±0.60 µmol m -2 s -1 ), g m 270 (0.21±0.02 mol m -2 s -1 ) and g m /g s (1.09 ±0.1) (Fig. 4a, 4b and  mol m -2 s -1 ), compared to control treatment (Fig. 5b). Increasing Cu concentration in soil 279 did not affect g m (Fig. 5c). The g m /g s ratio was greater under 20 and 75 µM Cu, 280 compared to control (Fig. 5d).  the same effect on g m , but to a lesser extent than g s . g m decreased from 0.27±0.02 to 296 0.19±0.02 mol m -2 s -1 under soil moisture deficit (Fig. 7c). In addition, g m /g s ratio 297 increased by 37% when plants were subject to a drought (Fig. 7d).

298
Relationship between CO 2 diffusion and photosynthetic activity 299 A max was strongly correlated to both g s and g m (P=0.001) and to V cmax (P=0.001) over all 300 the studies (Fig. 8a, 8b and 8c). Based on the collected data, g m was significantly 301 correlated to g s (P = 0.04). However, the relationship was not linear. g m was the highest compared to shade conditions Cano et al., 2013). 329 We observed a significant inverse relationship between g m and specific leaf area (SLA), AQPs after fertilization (Miyazawa et al., 2008b;Zhu et al., 2020).  (Miyazawa et al., 2008a).

431
Adaptation to the local environment might be a key driver of g m variation among taxa, 432 similarly to other morpho-physiological traits. Interspecific and intraspecific differences 433 in g m from mesic versus xeric environments (Quercus spp. and Eucalyptus spp.) were 434 reported by Zhou et al. (2014). Their study showed that g m , as well as g s , V cmax and J of 435 species from drier regions were less sensitive to water deficit which maintains 436 transpiration and photosynthesis activity at higher rates under drought, compared to 437 species from the mesic environment. Marchi et al. (2008)

Conflict of interests 470
The authors declare that there is no conflict of interest.

472
The datasets generated for this study are available on request to the corresponding 473 author.     ; mesophyll conductance (g m , c); g m /g s ratio (d); maximum rate of carboxylation (V cmax , e) and specific leaf area (SLA, f). In g m /g s ratio, g s for water (mol H 2 O m -2 s -1 ) was divided by 1.6 to obtain g s in mol CO 2 m -2 s -1 . Means having the same letters are not significantly different at α = 0.05. Figure 4. Effect of the ambient air CO 2 concentration on light-saturated photosynthetic rate (A max , a); stomatal conductance (g s , b); mesophyll conductance (g m , c) and g m /g s ratio (d). In g m /g s ratio, g s for water (mol H 2 O m -2 s -1 ) was divided by 1.6 to obtain g s in mol CO 2 m -2 s -1 . Means having the same letters are not significantly different at α = 0.05. Figure 5. Effect of the soil Copper (Cu) concentration on light-saturated photosynthetic rate (A max , a); stomatal conductance (g s , b); mesophyll conductance (g m , c) and g m /g s ratio (d). In g m /g s ratio, g s for water (mol H 2 O m -2 s -1 ) was divided by 1.6 to obtain g s in mol CO 2 m -2 s -1 . Means having the same letters are not significantly different at α = 0.05. Figure 6. Effect of the soil nitrogen content (high nitogen: HN, low nitrogen: LN) on light-saturated photosynthetic rate (A max , a); stomatal conductance (g s , b); mesophyll conductance (g m , c) and g m /g s ratio (d). In g m /g s ratio, g s for water (mol H 2 O m -2 s -1 ) was divided by 1.6 to obtain g s in mol CO 2 m -2 s -1 Means having the same letters are not significantly different at α = 0.05. Figure 7. Effect of the soil moisture on light-saturated photosynthetic rate (A max , a); stomatal conductance (g s , b); mesophyll conductance (g m , c) and g m /g s ratio (d). In g m /g s ratio, g s for water (mol H 2 O m -2 s -1 ) was divided by 1.6 to obtain g s in mol CO 2 m -2 s -1 . Means having the same letters are not significantly different at α = 0.05.

Figure 8.
Relationship between light-saturated photosynthetic rate (A max ), stomatal conductance (g s ), mesophyll conductance (g m ), maximum rate of carboxylation (V cmax ), electron transport rate (J), g m /g s ratio, specific leaf area (SLA) and per area leaf Nitrogen concentration (N area ). In g m /g s ratio, g s for water (mol H 2 O m -2 s -1 ) was divided by 1.6 to obtain g s in mol CO 2 m -2 s -1 .   ; mesophyll conductance (g m , c); g m /g s ratio (d); maximum rate of carboxylation (V cmax , e) and specific leaf area (SLA, f). In g m /g s ratio, g s for water (mol H 2 O m -2 s -1 ) was divided by 1.6 to obtain g s in mol CO 2 m -2 s -1 . Means having the same letters are not significantly different at α = 0.05. Figure 4. Effect of the ambient air CO 2 concentration on light-saturated photosynthetic rate (A max , a); stomatal conductance (g s , b); mesophyll conductance (g m , c) and g m /g s ratio (d). In g m /g s ratio, g s for water (mol H 2 O m -2 s -1 ) was divided by 1.6 to obtain g s in mol CO 2 m -2 s -1 . Means having the same letters are not significantly different at α = 0.05. Figure 5. Effect of the soil Copper (Cu) concentration on light-saturated photosynthetic rate (A max , a); stomatal conductance (g s , b); mesophyll conductance (g m , c) and g m /g s ratio (d). In g m /g s ratio, g s for water (mol H 2 O m -2 s -1 ) was divided by 1.6 to obtain g s in mol CO 2 m -2 s -1 . Means having the same letters are not significantly different at α = 0.05. Figure 6. Effect of the soil nitrogen content (high nitogen: HN, low nitrogen: LN) on light-saturated photosynthetic rate (A max , a); stomatal conductance (g s , b); mesophyll conductance (g m , c) and g m /g s ratio (d). In g m /g s ratio, g s for water (mol H 2 O m -2 s -1 ) was divided by 1.6 to obtain g s in mol CO 2 m -2 s -1 Means having the same letters are not significantly different at α = 0.05. Figure 7. Effect of the soil moisture on light-saturated photosynthetic rate (A max , a); stomatal conductance (g s , b); mesophyll conductance (g m , c) and g m /g s ratio (d). In g m /g s ratio, g s for water (mol H 2 O m -2 s -1 ) was divided by 1.6 to obtain g s in mol CO 2 m -2 s -1 . Means having the same letters are not significantly different at α = 0.05. . Relationship between light-saturated photosynthetic rate (A max ), stomatal conductance (g s ), mesophyll conductance (g m ), maximum rate of carboxylation (V cmax ), electron transport rate (J), g m /g s ratio, specific leaf area (SLA) and per area leaf Nitrogen concentration (N area ). In g m /g s ratio, g s for water (mol H 2 O m -2 s -1 ) was divided by 1.6 to obtain g s in mol CO 2 m -2 s -1 .