Introduction

Conifer forests thrived on the Earth surface during the Mesozoic period until the radiation and diversification of angiosperms during the Cretaceous, which was followed by angiosperm ecological dominance attributed to increased physiological performance and reduced generation times (Bond, 1989; Boyce et al., 2009; Crepet & Niklas, 2009; de Boer et al., 2012). Yet, after 100 Myr of competition with the angiosperms, conifers remain prominent in the canopy of many biomes (Brodribb et al., 2012). Conifers are found in ecosystems from high latitudes to the equator, and they have major economic importance in the wood and paper industries (McFarlane & Sands, 2013). Conifer-dominated forests are not exempt from the impacts of drought and aridity resulting from ongoing global climatic changes (Dai, 2011; Brodribb et al., 2020; Kharuk et al., 2021). This is particularly alarming given that 48% of the 722 conifer taxa of the world are currently under threat (Blackmore et al., 2011). Global increases in temperature coupled with rising vapour pressure deficit (VPD) place increased strain on plant hydraulic and photosynthetic systems (Choat et al., 2018; Grossiord et al., 2020). There is strong evidence that tree water use efficiency (WUE) has increased in recent decades, most probably the result of rising atmospheric CO₂, allowing plants to open their stomata less frequently, thereby conserving water (Keenan et al., 2013; Mathias & Thomas, 2021). However, the underlying physiological mechanisms behind this trend need to be further elucidated (Guerrieri et al., 2019). Furthermore, there is still a lack of knowledge on how leaf structural organisation influences key functions such as photosynthetic carbon acquisition, stomatal conductance, and the interplay between both driving intrinsic water use efficiency (WUE_i).

Leaf-level photosynthetic metabolism has an important role in maintaining global ecological processes (Hetherington & Woodward, 2003). Therefore, exploring tissue organisation inside the leaf, and specifically the mesophyll cells where water and gas diffusion occur, is important for understanding carbon, water and energy fluxes at whole ecosystem levels. The leaf mesophyll consists of photosynthetic parenchyma cells located between the epidermis and the endoderm and transfusion tissue layers that surround the veins. Once inside the leaf, diffusion of CO₂ from the intercellular airspace (IAS) to the chloroplast stroma (i.e. mesophyll conductance) is a major constraint on photosynthetic performance (Flexas et al., 2012; Gago et al., 2020). This pathway includes the IAS, but also the diffusion of CO₂ across mesophyll cell walls, cell membranes and the chloroplast envelope, which can significantly limit photosynthesis in gymnosperms (Carriquí et al., 2020). Previous work has shown that mesophyll surface area per leaf area (S_m; µm² µm⁻²) has a strong influence on maximum photosynthesis (Nobel et al., 1975; Smith & Nobel, 1977). More recently, it has been suggested that the surface area of the mesophyll exposed to the IAS per unit of mesophyll volume (SA_mes/V_mes; µm² µm⁻³) can influence plant photosynthetic performance given that the mesophyll-IAS boundary is the primary interface between the atmosphere and the photosynthetic cells (Earles et al., 2018; Théroux-Rancourt et al., 2021). Other volumetric anatomical traits such as mesophyll porosity (i.e. IAS as a fraction of total mesophyll volume; µm³ µm⁻³) might be relevant in promoting diffusion through the IAS, and it has been suggested that mesophyll palisade porosity is correlated with stomatal conductance across four different Arabidopsis mutants (Lundgren et al., 2019). The relevance of such anatomical traits arises from the hypothesis that photosynthetic capacity could be enhanced by increasing surface area of mesophyll exposed to the IAS, allowing for more potential surface for CO₂ diffusion across mesophyll cell walls, and chloroplast envelopes (Evans et al., 2009; Earles et al., 2018). However, the correct estimation of such anatomical traits using standard two-dimensional (2D) techniques is difficult as it relies on 2D approximations of the complex, three-dimensional (3D) shape of the mesophyll (Théroux-Rancourt et al., 2017; Earles et al., 2019).

Given the elementary vascular architecture of the conifer leaf, which typically consists of a single vein without further branching (Zwieniecki et al., 2004; Brodribb et al., 2010), and the large variation in conifer leaf shape (Fig. 1), area-based traits might not allow for an accurate comparison of strategies to optimise leaf structure with function within conifers (Earles et al., 2019; Théroux-Rancourt et al., 2021). Furthermore, given the structural relevance of the central vasculature, traits other than those related to the mesophyll might also influence CO₂ and water diffusion in the conifer leaf. Therefore, a central question is whether nonvascular tissues are tightly coupled to vein volume with proportional relationships, or whether they show greater variability to compensate for reduced hydraulic efficiency. We predict that higher volumes of vascular tissue per unit leaf volume (V_vein/V_leaf; µm³ µm⁻³) should be a good predictor of gas exchange and WUE_i because of the mechanistic link between hydraulic conductance and maximum photosynthetic rates in conifers (Brodribb & Feild, 2000). We also predict a positive relationship between gas exchange efficiency and the ratio of mesophyll surface area exposed to the IAS and vein volume (SA_mes/V_vein; µm² µm⁻³), where a greater investment in vein volume per unit area of bulk tissue surface should increase the hydraulic capacity to replace water lost to transpiration. In conifer leaves, water moves across the transfusion tissue and a ring of parenchymatous sheath cells or an endoderm before reaching the mesophyll (Hu & Yao, 1981). Therefore, transfusion tissue and endoderm volume relative to total leaf volume (V_TT+en/V_leaf; µm³ µm⁻³), should also influence the efficiency of water movement within the leaf. Along with the previously described features, many conifer species also possess resin ducts, which play a major role in chemical and physical defence (Gaylord et al., 2013; Breshears et al., 2018), but necessarily displace vascular or photosynthetic tissue, incurring both maintenance and construction costs, but also lost opportunity costs for net carbon gain. Finally, water movement inside the leaf ends at the stomatal pores, which play a major role in regulating water loss and maintaining water status in conifers (Brodribb et al., 2014). Within this context, it is possible to determine the proportionality of different anatomical traits, the coordination between supply and demand for CO₂ and H₂O, and the physical constraints of leaf construction. To probe these relationships, we describe how conifers build their elementary, yet diverse leaves, and how structural features relate to key physiological traits.

This study presents a survey of the three-dimensional organisation of the conifer leaf using microtomography (microCT) imaging (Supporting Information Table S1). Our study includes 34 conifer species (Fig. S1), with a special focus on the genus Pinus, the largest extant genus of conifers (Gernandt et al., 2005). Pinus species can be found in a broad range of environmental conditions, suggesting wide structural and functional diversity (Martínez-Vilalta et al., 2004). Pinus subgenera Pinus and Strobus can be distinguished by having two or one vascular bundles respectively, centrally located inside the needle-like leaf (Gernandt et al., 2005) (Fig. 1a–c). Despite their rather simple anatomical organisation, conifer leaves have extensive morphological diversity ranging from flat leaves to needle-like leaves with different degrees of transversal flattening. Leaf morphological diversity in conifers results in different physiological performances, with flat-leaved species having lower photosynthetic assimilation and respiration rates than needle-leaved species (Brodribb & Feild, 2008; Schmiege et al., 2020, 2021). Yet, the differences in 3D anatomical structure across conifers with different leaf morphologies, which could explain their contrasting photosynthetic performance, need to be elucidated. We hypothesise that features enhancing mesophyll surface area for gas diffusion, will be positively correlated with the light-saturated assimilation rate of CO₂ (A_sat) and stomatal conductance (g_s). We also provide a volume-based stomatal density estimation, a trait we expect better captures the interplay between the evaporative surfaces invested in the nonlaminar mesophyll volume and the number of stomata needed to provide CO₂, with the expectation that leaves with higher number of stomata per mesophyll volume would have both higher rates of gas exchange and WUE_i due to an enhancement of the epidermal pores serving as evaporative surface relative to the photosynthetic tissue.

Materials and Methods

Results

The mesophyll (V_mes/V_leaf), including cells and airspace, represents the dominant leaf volume fraction for all 34 measured conifer species in this study, occupying an average of 60% of the total leaf volume (Figs 1, S2). The second largest volume fraction was either the combined transfusion tissue and endoderm (V_TT+en/V_leaf) that surrounds the vascular cylinder, or the epidermis (V_ep/V_leaf), which represented an average of 22% and 16% of the leaf volume, respectively (Fig. 1a). Veins (V_vein/V_leaf) and resin ducts (V_resin/V_leaf) represented the smallest fraction of the total leaf volume (Fig. 1a), and resin ducts were completely absent in six measured species (Dataset S1). We found weak structural coordination amongst tissue volumes within the conifer leaf (SMA; Table S2). However, a negative allometric scaling between V_ep/V_leaf and V_mes/V_leaf was observed (r² = 0.27, P < 0.01), indicating that increasing the relative allocation to the mesophyll was done in conjunction with a decrease in the relative allocation to the epidermis. Average values of all measured volumetric traits for each species are included in Dataset S1.

A multivariate analysis of trait covariation defined two major axes explaining 48% of inertia (Fig. 2). Inertia of the first axis was mainly explained by V_IAS/stomate, the amount of air volume connected to a stomate (18.27%), stomata/V_mes, the stomatal density per mesophyll volume (15.04%), V_mes/V_leaf (14.74%) and g_s (14.08%). Increasing g_s was associated on the first axis with increases in stomata/V_mes and A_sat, and decreases in mesophyll porosity, V_mes/V_leaf, and V_IAS/stomate (Fig. 2). The second axis was largely explained by V_vein/V_leaf (18.04%), SA_mes/V_vein (14.83%) and S_m (14.45%). Increasing WUE_i was associated on the second axis with increases in S_m and V_vein/V_leaf, and decreases in stomata/V_mes (Fig. 2). Leaf morphological types were dispersed along both major axes (Fig. 2). Needle-like leaves were isolated due to their higher V_ep/V_leaf and stomata/V_mes. Flat leaves and flattened needles largely overlapped due to their porous and voluminous mesophylls as opposed to needle-like leaves (Fig. 3a,b). Flat leaves and flattened needles also converged in having lower stomata/V_mes (Fig. 3c) and higher V_IAS/stomate (Fig. 3d) than needle-like leaves. However, flattened needles were differentiated due to some species having significantly higher SA_mes/V_vein (Fig. 2; Table S3). Conifer groups were also effectively segregated based on their volumetric anatomy. Non-Pinus conifer species were segregated on the first axis (Fig. 2) due to their higher porosity (Fig. 1d), V_mes/V_leaf, and V_IAS/stomate values, while species from the Pinus subgenus were grouped (Fig. 2) due to higher V_ep/V_leaf (Fig. 1b), along with a relatively larger SA_mes/V_mes. Species from the Strobus subgenera were located between the previously described conifer groups (Fig. 2). Structural divergences of leaf morphologies, distinguishing three distinct functional groups, along with the segregation of different conifer clades in the PCA analysis were further supported with one-way ANOVA analyses on 3D leaf anatomical traits and stomatal density (Tables S3, S4). Relative vein volume and SA_mes/V_mes were highly conserved (Fig. S3; Tables S3, S4), whereas SA_mes/V_vein, along with traits related to the ratio of stomatal pore number and mesophyll tissue volumes had significant differences across both leaf morphologies and conifer groups (Figs 3c,d, S3; Tables S3, S4).

To explore the functional implications of 3D tissue content, we determined their relationships to gas exchange parameters such as A_sat, g_s and WUE_i. WUE_i was best predicted by V_vein/V_leaf (Fig. 4a), where higher V_vein/V_leaf enhances leaf WUE_i. Furthermore, 2D anatomical estimations of the ratio A_vein/A_leaf were comparable with V_vein/V_leaf, extracted using a 3D approach (Notes S1; Fig. S4). The mesophyll surface area exposed to vein volume (SA_mes/V_vein) was also an accurate predictor and showed a negative relationship with WUEi (Fig. 4b). A positive relationship of WUE_i with S_m was also found using generalised least-square models corrected for phylogenetic relatedness (PGLS) analyses (Table S5). A_sat was the physiological trait showing the least linkage with 3D structural traits. However, we found a negative relationship between A_sat and mesophyll porosity along with V_mes/V_leaf (Fig. S5a), suggesting that conifer species with lower relative mesophyll volumes have greater photosynthetic assimilation rates. Additionally, V_IAS/stomate was also negatively related to A_sat (Fig. S5b). The number of stomata per unit mesophyll tissue volume predicted g_s and WUE_i (Fig. 5). For instance, high stomatal densities relative to mesophyll volume (stomata/V_mes) enhanced g_s (Fig. 5a) while decreasing WUE_i (Fig. 5b). Evolutionary coordination between leaf volumetric anatomy and physiological traits was supported by PGLS analyses (Table S5). Stomatal density measured on a 2D leaf surface area basis (stomata/A_leaf), included here as a reference of a more standard approach, did not relate to any physiological trait (Table S5). Therefore, the interaction of mesophyll volume and stomatal pore numbers emerged as a key trait to explain leaf physiological performance (Fig. 5). A comparison of the studied conifers with published data for other gymnosperm species, along with angiosperms and ferns, showed that conifers had fewer stomata per unit mesophyll volume than angiosperms sensu largo (Fig. S6; Dataset S3). However, this difference was less important when considering evergreen angiosperms alone. Conifers showed a similar stomata/V_mes ratio as other gymnosperm species and ferns (Fig. S6).

Discussion

Conifers, with their long-lasting and anatomically elementary leaves, must rely on leaf construction to enable sufficient carbon assimilation to survive and reproduce while limiting water loss. However, the simple design of coniferous leaves has been considered to have a poor, if not absent, hydraulic connection between the vein xylem and the bulk leaf tissue (Zwieniecki et al., 2007). This is unsurprising as conifer leaves are not as fully vascularised as those of angiosperms and present a single cohort of relatively inefficient leaves (Bond, 1989). The likely consequence of this poor hydraulic connection is a larger difference in water potential between the veins and the mesophyll and epidermis in transpiring leaves, which may force stomata to close even when water potential is relatively high in the veins (Zwieniecki et al., 2007). Our results show that, despite occupying a small fraction (c. 2%) of the leaf volumetric matrix, vein tissue volume has a great impact on the leaf WUE_i (Fig. 4a). As the relative vein volume expands, the space between the vascular tissues and the bulk leaf tissues become smaller, potentially reducing the hydraulic resistance for water transport from the vasculature to the bulk leaf. Our study provides an estimation of the full volumetric space occupied by the vein tissue relative to the photosynthetic cells. While we consider volumetric estimates to be more accurate given that they integrate traits over a larger leaf fraction than 2D estimates, we found that standard 2D anatomical estimations of the relative surface covered by veins over a few leaf cross sections could be accurately used to predict the vein volumetric fraction and, in turn, WUE in conifer leaves (Notes S1; Fig. S4). Stomatal and venation densities covary in angiosperms, directly affecting gas exchange and water use capacities (Carins Murphy et al., 2014). However, conifers do not provide a dense network of veins within the mesophyll to irrigate the photosynthesising cells in water. Consequently, conifers might be constrained to increase the relative volume of veins as a single way to provide more water to the leaf.

Previous work has suggested that narrow, needle-like leaves in conifers, which is a common feature in the Pinaceae (Brodribb & Feild, 2008; Brodribb et al., 2010), would be a response to alleviate their lack of hydraulic redundancy. With the photosynthetic tissue encircling the single vascular cylinder, the distance from the vein to the epidermis largely sets hydraulic conductance outside the xylem, and mesophyll volume becomes a major limiting factor in cylindrical, narrow leaves. Increasing the radial pathlength for water transport should effectively limit leaf hydraulic conductance, in turn limiting photosynthetic rates (Brodribb et al., 2010). Our data support this hypothesis, with a negative relationship between mesophyll volume and A_sat (Fig. S5a). Therefore, within our dataset we found evidence for a significant photosynthetic penalty for increasing leaf width and mesophyll volume in flat and flattened needle leaves (Fig. 3b). With a limited ability to maximise photosynthetic capacity through hydraulic ramification, the cylindrical needle-like leaf may offer opportunities for other structural elements that allow improved hydraulic performance. Accessory transfusion tissue is one of them, where specialised cells connect the veins to the mesophyll tissue, providing more water to the photosynthesising cells and improving hydraulic contact to the epidermis (Brodribb et al., 2007). This might lead to increased photosynthetic rates in needle-like leaves, which presented the most voluminous transfusion tissues (Table S3). Our results support this linkage, with a positive relationship of the combined volumes of transfusion tissue and endoderm with photosynthetic assimilation (Table S5). It has been stated that reaching high photosynthetic rates requires high leaf porosity values, which might increase CO₂ diffusion (Brodribb et al., 2020). Yet, we observed a different trend with a negative relationship between mesophyll porosity and A_sat (Table S5). Moreover, A_sat was strongly negatively related with V_mes/V_leaf in our dataset (Fig. S5a). Illumination-induced fluorescence has been shown to be more evenly distributed in needle-like leaves lacking palisade mesophyll layers (Johnson et al., 2005). Therefore, the decrease of A_sat in flat leaves with more voluminous mesophylls and often presenting a more differentiated palisade layer might be explained by a limitation of light propagation across the mesophyll. Interestingly, we found conspicuous differences in leaf design between both Pinus subgenera and other conifers (Figs 1, 2; Table S4), with narrow needle-leaved Pinus possessing less voluminous and porous mesophylls. Such differences in mesophyll construction, which enhance the exchange surface by condensing the mesophyll cells in the more compact mesophyll of needle-like leaves, could explain why A_sat is greatest in Pinus species.

SA_mes/V_vein is another feature that has diverged across leaf morphologies and conifer clades (Tables S3, S4). We propose SA_mes/V_vein as another anatomical trait involved in regulating the control over the loss of water (Fig. 4b). Minimising this ratio would mean that less surface is available for evaporation for a given vein water volume, increasing the time before this capacitor is depleted, and thereby lowering the ‘safety margin’ between stomatal closure and xylem cavitation (Zwieniecki et al., 2007). Also, in the context of poorly connected hydraulic design, decreasing SA_mes/V_vein would minimise the apoplastic surface for water to travel from the vein to the epidermis, therefore by proxy decreasing the water path length and increasing connectivity to the epidermis to allow stomata to stay open longer and photosynthesis to continue. In our dataset, decreasing SA_mes/V_vein was achieved mainly by increasing the relative volume of veins, that is investing more in vascular tissue (Fig. 4). The relative benefit of investing in vein volume to increase the efficiency in water use seems to plateau above c. 3% of leaf volume, and most species invest in veins close to that relative volume (Fig. 4a). Pinus species, which bear needle-like leaves have, in addition, decreased the relative volume of the mesophyll by decreasing airspace volume (Figs 1, 3a,b, S3a,b) with plicate mesophyll cells (Esau, 1977). Decreasing porosity leads to more cells being in contact with each other, thereby decreasing SA_mes, that is the surface of cells exposed to the IAS. Increasing IAS had a positive effect on WUE in six angiosperm species (Mediavilla et al., 2001). Although positive, we could not find a significant relationship between mesophyll porosity and WUE_i in conifer leaves (Table S5). Beyond considering mesophyll features alone, mesophyll volumetrics in interaction with stomatal pore number emerged here as key traits to explain conifer gas exchange and WUE_i (Fig. 5). Using Arabidopsis and wheat as model plants, it has been suggested that stomatal differentiation during leaf development might induce mesophyll airspace formation (Lundgren et al., 2019). Our study shows that this coordination between V_IAS, along with V_mes, and stomatal number have a significant impact on carbon assimilation and gas exchange on conifers (Fig. 5; Table S5), further supporting this crucial linkage. Considered in a wider context, our observations might provide a novel structural basis to explain the lower photosynthetic rates of ferns and gymnosperms compared with angiosperms, as we show that they have greater mesophyll volume per stoma, acting as a bottleneck that limits their evaporative capacities (Fig. S6).

Current increases in temperature and atmospheric CO₂ concentrations might impact the structure and function of conifer forests worldwide, and it has been proposed that improved WUE could alleviate the temperature effect (Brodribb et al., 2020). Previous studies have shown higher WUE_i under increasing CO₂ in conifer species, having stronger WUE_i responses than angiosperms (Dalling et al., 2016; Klein et al., 2016; Adams et al., 2020). Additionally, it has been recently shown that higher plasticity in the vascular tissue of the needles of Pinus pinaster enhances their WUE_i (Bert et al., 2021). Using an experimental approach, needles of Larix kaempferi growing under higher CO₂ showed increased mesophyll surface area per leaf area, coupled with higher photosynthetic rates (Eguchi et al., 2004). Moreover, it was shown that elevated CO₂ increased mesophyll surface and decreased stomatal density in Pinus sylvestris needles (Lin et al., 2001). Therefore, under such elevated CO₂ conditions, we might expect to observe a lower stomata/V_mes ratio, which would enhance WUE_i according to our predictions (Fig. 5b). Furthermore, given the recently demonstrated link between enhanced WUE_i and vascular tissue plasticity in conifer needles (Bert et al., 2021), we expect that coordinated changes in vascular and mesophyll tissue volumetrics, along with shifts in stomatal pore number in conifer leaves, may allow conifer species to cope and adapt to the pressure exerted by increasing VPD in many global biomes (Grossiord et al., 2020) by maintaining similar carbon assimilation levels with lower water consumption.

Author contributions

CB, DMJ, GT-R, ST and TNB designed research; CB, DL, DMJ, GT-R, JME and ST performed measurements and collected data; ST analysed the data and wrote the first draft of the manuscript with major contributions from CB and GT-R. All authors contributed to manuscript revisions.