Mesophyll conductance to CO2: current knowledge and future prospects



    Corresponding author
    1. Grup de Recerca en Biologia de les Plantes en Condicions Mediterrànies, Departament de Biologia, Universitat de les Illes Balears, Carretera de Valldemossa Km 7.5, 07122 Palma de Mallorca, Balears, Spain and
      J. Flexas. Fax: +34 971 173184; e-mail:
    Search for more papers by this author

    1. Grup de Recerca en Biologia de les Plantes en Condicions Mediterrànies, Departament de Biologia, Universitat de les Illes Balears, Carretera de Valldemossa Km 7.5, 07122 Palma de Mallorca, Balears, Spain and
    Search for more papers by this author

    1. Instituto de Recursos Naturales y Agrobiología, CSIC, Apartado 1052, 41080 Sevilla, Spain
    Search for more papers by this author

    1. Grup de Recerca en Biologia de les Plantes en Condicions Mediterrànies, Departament de Biologia, Universitat de les Illes Balears, Carretera de Valldemossa Km 7.5, 07122 Palma de Mallorca, Balears, Spain and
    Search for more papers by this author

    1. Grup de Recerca en Biologia de les Plantes en Condicions Mediterrànies, Departament de Biologia, Universitat de les Illes Balears, Carretera de Valldemossa Km 7.5, 07122 Palma de Mallorca, Balears, Spain and
    Search for more papers by this author

J. Flexas. Fax: +34 971 173184; e-mail:


During photosynthesis, CO2 moves from the atmosphere (Ca) surrounding the leaf to the sub-stomatal internal cavities (Ci) through stomata, and from there to the site of carboxylation inside the chloroplast stroma (Cc) through the leaf mesophyll. The latter CO2 diffusion component is called mesophyll conductance (gm), and can be divided in at least three components, that is, conductance through intercellular air spaces (gias), through cell wall (gw) and through the liquid phase inside cells (gliq). A large body of evidence has accumulated in the past two decades indicating that gm is sufficiently small as to significantly decrease Cc relative to Ci, therefore limiting photosynthesis. Moreover, gm is not constant, and it changes among species and in response to environmental factors. In addition, there is now evidence that gliq and, in some cases, gw, are the main determinants of gm. Mesophyll conductance is very dynamic, changing in response to environmental variables as rapid or even faster than stomatal conductance (i.e. within seconds to minutes). A revision of current knowledge on gm is presented. Firstly, a historical perspective is given, highlighting the founding works and methods, followed by a re-examination of the range of variation of gm among plant species and functional groups, and a revision of the responses of gm to different external (biotic and abiotic) and internal (developmental, structural and metabolic) factors. The possible physiological bases for gm, including aquaporins and carbonic anhydrases, are discussed. Possible ecological implications for variable gm are indicated, and the errors induced by neglecting gm when interpreting photosynthesis and carbon isotope discrimination models are highlighted. Finally, a series of research priorities for the near future are proposed.


During photosynthesis, CO2 has to move from the atmosphere surrounding the leaf across a boundary layer in the air above the foliage surface to the sub-stomatal internal cavities through the stomata (Fig. 1a), and from there to the site of carboxylation inside the stroma through the leaf mesophyll (Fig. 1b). From Fick's first law of diffusion, the net photosynthetic flux at steady state (AN) can be expressed as: AN = gs (Ca − Ci) = gm (Ci − Cc), where gs and gm are the stomatal and mesophyll conductance to CO2 diffusion, and Ca, Ci and Cc are the CO2 concentrations in the atmosphere, in the sub-stomatal internal cavity and in the chloroplast stroma, respectively (Long & Bernacchi 2003). While this two-dimensional view of CO2 diffusion is mostly valid when referring to the stomata, it is rather simplistic when referring to the mesophyll, and it may be inadequate to describe a system where there are CO2 sinks and sources distributed along the pathway (Parkhurst 1994). However, although there have been attempts to develop three-dimensional models for CO2 mesophyll diffusion (Parkhurst 1994; Aalto & Juurola 2002; Juurola et al. 2005), at present, they are far from being of general use. Nevertheless, as pointed out by Parkhurst (1994), ‘representing internal diffusion limitations with single numbers seems preferable to neglecting them completely’. Therefore, the present review intends to update the current knowledge on mesophyll conductance from a two-dimensional perspective, considering only vertical CO2 diffusion through leaf mesophyll. Lateral CO2 diffusion inside the mesophyll can also be important for photosynthesis in some cases (Morison et al. 2005; Pieruschka, Schurr & Jahnke 2005), although this aspect has already been reviewed recently (Morison & Lawson 2007).

Figure 1.

(a) Micrograph of the abaxial surface of an olive leaf (bottom side up), where the stomata can be seen, as well as the pathway of CO2 from ambient (Ca) through leaf surface (Cs) and intercellular air spaces (Ci) to the chloroplast (Cc). Boundary layer conductance (gb), stomatal conductance (gs) and mesophyll conductance (gm) are indicated. (b) Electron micrograph of a grapevine leaf where cell wall (cw), plasma membrane (pm), the chloroplast envelope (ce) and stroma thylakoid (st) can be observed. The pathway of CO2 from Ci to chloroplastic CO2 (Cc) is characterized by intercellular air space conductance to CO2 (gias), through cell wall (gw) and through the liquid phase inside the cell (gliq). A grain of starch (s) and a plastoglobule (pg) can be also observed in the picture (photos by A. Diaz-Espejo).

In Gaastra's (1959) pioneer work on leaf photosynthesis, mesophyll conductance (and its inverse, mesophyll resistance) was defined essentially as a diffusion component of the photosynthesis pathway, and it was considered to be an important factor in determining leaf photosynthesis (actually more important than stomatal conductance). Later, Troughton & Slatyer (1969) extended the use of the term to a ‘mixed’ diffusion-biochemical component, that is, to refer to the initial slope of AN versus Ci relationship. Following Gaastra's (1959) assumption that CO2 concentration inside the chloroplast was near zero (i.e. at the CO2 compensation concentration), Jones & Slatyer (1972b) proposed a method to separate the transport and carboxylation components of what continued to be called mesophyll or intracellular conductance. Using this method, earlier conclusions by Gaastra that mesophyll transport resistance was a limiting factor for photosynthesis were confirmed (Jones & Slatyer 1972a,b; Samsuddin & Impens 1979), and it was suggested, for the first time, that the leaf internal resistance to CO2 transfer could be variable and could respond, for instance, to water stress (Jones 1973). The nomenclature was further complicated by Samsuddin & Impens (1979), who termed internal resistance to the ‘mixed’ diffusion-biochemical term, while restoring the use of the term mesophyll resistance to refer to the transfer component alone. Such a semantic confusion continues nowadays, and several more recent papers still call ‘mesophyll conductance’ or ‘internal conductance’ to the initial slope of AN versus Ci relationship (e.g. Moldau & Kull 1993; Laisk & Loreto 1996; Eichelmann et al. 2004a,b; Béjaoui et al. 2006). We propose, as already did Farquhar & Sharkey (1982) and Parkhurst (1994), to restrict the term mesophyll conductance to the diffusion of CO2 through leaf mesophyll, both through intercellular air spaces (gias), cell wall (gw) and the intracellular liquid pathway (gliq). That is, mesophyll conductance should be viewed as synonymous of ‘leaf internal diffusion conductance’, while terms such as ‘apparent carboxylation efficiency’ or ‘apparent carboxylation conductance’ may be more adequate to refer to the initial slope of the AN versus Ci relationship.

With the introduction of the most commonly used leaf photosynthesis model by Farquhar, von Caemmerer & Berry (1980), the early assumption that CO2 concentration in the chloroplasts was close to zero or to the compensation point was rejected, as was later being confirmed by direct measurements of Ci (Lauer & Boyer 1992). Thereafter, most gas exchange studies have usually assumed that gm is large and constant, that is, that Ci = Cc. However, different studies in the late 80s and early 90s have already suggested that Cc is significantly less than Ci, although not close to zero. Evans et al. (1986) reached this conclusion by comparing online carbon isotope discrimination of photosynthesizing leaves with the theoretical discrimination that would be expected if Ci = Cc. Evans (1983) and Evans & Terashima (1988) reached the same conclusion by comparing the initial slope of the AN versus Ci curve in wheat and spinach leaves with the activity of ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco) as determined in vitro. With the advent of pulse-amplitude fluorometers, the comparison of chlorophyll fluorescence with gas exchange measurements also indicated that Cc was lower than Ci, and that their difference increased under conditions of water stress or salinity (Bongi & Loreto 1989; Cornic et al. 1989; Di Marco et al. 1990). The water stress-induced decrease of Cc below Ci was independently demonstrated by measuring leaf 18O discrimination (Renou et al. 1990), and was lately confirmed several times by both chlorophyll fluorescence and leaf 18O discrimination (Tourneux & Peltier 1995; Sánchez-Rodríguez, Pérez & Martínez-Carrasco 1999).


The pioneer studies described in the previous section includes the development of several methods that have become usual to estimate mesophyll conductance to CO2. A detailed description of the principles and procedures of these methods has recently been reviewed by Warren (2006a). Therefore, the present review will only mention the existing methods, providing the original and extended references for readers interested in a more detailed description.

Currently, widely used methods include (1) the single-point online carbon isotope discrimination method introduced by Evans et al. (1986) and further developed by Sharkey et al. (1991) and Loreto et al. (1992), (2) the slope-based online carbon isotope discrimination method also introduced by Evans et al. (1986) and further developed by von Caemmerer & Evans (1991) and Lloyd et al. (1992), (3) the ‘constant J’ fluorescence method introduced by Bongi & Loreto (1989) and further developed by Harley et al. (1992), and (4) the ‘variable J’ fluorescence method introduced by Di Marco et al. (1990) and also further developed by Harley et al. (1992). Two variations of the latter method have also been described, namely, the ‘variable J’ fluorescence method by Epron et al. (1995) and the ‘J at the CO2 photocompensation point’ method by Laisk et al. (2002). Alternative methods include (5) the ‘initial slope’ method by Evans (1983) and Evans & Terashima (1988), (6) the carbon isotope discrimination in recently synthesized sugars originally described by Brugnoli & Lauteri (1991) and further developed by Lauteri et al. (1997), and (7) the ‘real versus apparent compensation point (Γ* versus Ci*)’ method by Peisker & Apel (2001). Recently, a novel (8) ‘curve-fitting’ method was introduced by Ethier & Livingston (2004), which was followed by a similar method by Sharkey et al. (2007). In addition, the gas phase component of gm (gias) can be determined by comparing leaf CO2 assimilation in atmospheres containing different gases, such as NO2 (Jarvis & Slatyer 1970) or helium (Farquhar & Raschke 1978; Parkhurst & Mott 1990; Genty et al. 1998; Piel 2002). On the other hand, the liquid phase component of gm (gliq) can be measured by using leaf 18O discrimination, as originally described by Renou et al. (1990) and further developed by Tourneux & Peltier (1995) and, specially, by Gillon & Yakir (2000), or by using the photoacoustic method described by Gorton, Herbert & Vogelmann (2003).

Each of these methods relies on a number of assumptions and requires the determination of several parameters that sometimes are not easy to obtain. Moreover, the recent demonstration that gm is largely and rapidly affected by CO2 and light intensity (Flexas et al. 2007a) impairs the fundamentals of some of these methods, including the ‘slope-based carbon isotope discrimination’ method, the constant J method and the curve-fitting method (although the curve-fitting methods give reliable estimates of the ‘average’gm over a given CO2 range). When comparing gm values among species, treatments, etc., it is important to keep in mind the limits of precision of these methods. Warren (2006a) reported that SDs could be as high as 19–38% of the mean for the chlorophyll fluorescence methods, and 6–34% for the isotopic method. Therefore, the methods available do not allow solving for small differences in gm. Nevertheless, the fact that comparing several of these methods, which rely in substantially different assumptions, yields similar results reinforces the idea that they all provide reasonable estimates of gm in leaves (Loreto et al. 1992; Flexas et al. 2006b, 2007a; Warren & Dreyer 2006).

Despite limits in their accuracy, the use of all these methods has increased our capacity to explore mesophyll conductance to CO2 and its relevance for photosynthesis. The number of references addressing several aspects of mesophyll conductance to CO2 is increasing exponentially, and up to about a hundred publications has appeared over the last 20 years containing estimations of gm and/or reviewing any aspect of its importance for plant physiology (Fig. 2). Actually, the number of reports on gm has doubled during the past 5 years (Fig. 2), because of the introduction early this decade of commercial devices capable of simultaneous gas exchange and chlorophyll fluorescence measurements over the same leaf area [Li-Cor Inc. (NE, USA) introduced the first of such systems in June 2001].

Figure 2.

Evolution of publications on mesophyll conductance to CO2 (gm) over the last 21 years. The number of published papers each year is shown as bars, while the accumulated number of papers is shown as a dotted trend. For 2007, only papers published or were in press by the end of August were considered. The papers considered include those describing methods for gm estimation, reviewing its importance in any aspect of plant ecophysiology or providing estimates of gm in plants.

The aim of the present review was to summarize the accumulated knowledge on mesophyll conductance to CO2 and to stress some future research needs. Previous reviews on CO2 diffusion inside leaves are available (Parkhurst 1994; Evans & von Caemmerer 1996; Evans & Loreto 2000; Massacci & Loreto 2001). In addition, other reviews have addressed specific aspects of gm, such as its importance for sun–shade differentiation (Terashima, Miyazawa & Hanba 2001; Niinemets & Sack 2005; Terashima et al. 2006), its role in photosynthetic nitrogen and water use efficiency (WUE) (Hikosaka 2004; Warren & Adams 2006), its importance for photosynthesis optimization in different plant groups (Warren & Adams 2004; Terashima et al. 2005) or the necessity of incorporating a term for gm in the parameterization of photosynthesis models (Long & Bernacchi 2003; Ethier & Livingston 2004; Sharkey et al. 2007).


It is now well established that mesophyll conductance to CO2 is finite and of similar magnitude as stomatal conductance (Evans & von Caemmerer 1996; Evans & Loreto 2000). Data are available in the literature for up to 122 different species, subspecies, hybrids, forms and varieties. When these data are averaged for different groups, differences in the maximum attainable gm (i.e. in the absence of stress conditions) are revealed (Fig. 3). In general, herbaceous annual and biannual plants present the largest values of gm (around 0.4 mol CO2 m−2 s−1 bar−1), with no differences between monocots and dicots (not shown). Perennial herbs and woody deciduous angiosperms display significantly lower values (around 0.2 mol CO2 m−2 s−1 bar−1), while the very few woody semi-deciduous angiosperms in which gm has been determined show intermediate values. In woody evergreen plants, gm is even lower (slightly above and below 0.1 mol CO2 m−2 s−1 bar−1 in angiosperms and gymnosperms, respectively). The very few data available for succulent CAM plants suggest that their gm is as low as that of gymnosperms (Maxwell, von Caemmerer & Evans 1997; Griffiths et al. 2007).

Figure 3.

Maximum mesophyll conductance values (i.e. in the absence of ambient stress conditions) for different plant groups, including woody evergreen gymnosperms, woody evergreen angiosperms, woody semi-deciduous angiosperms, woody deciduous angiosperms, herbaceous perennials, herbaceous annual/biannual plants and CAM plants. For herbaceous plants, monocots and dicots are plotted together because there were no differences in average values between the two groups. Values are average ± SE of published results. Data averaged include the following species: evergreen gymnosperms (n = 13): Abies concolor54, Abies grandis54, Abies magnifica54, Abies procera54, Dacrydium cupressinum41, Pinus lambertiana54, Pinus monticola54, Pinus pinaster75, Pinus radiata41, Picea sitchensis59, Prumnopitys ferruginea41, Pseudotsuga menziesii41,54,57,67, Tsuga heterophylla54; woody evergreen angiosperms (n = 32): Arbutus unedo7, Camellia japonica25,44, Castanopsis sieboldii25,30,44, Cinnamomum camphora25, Citrus aurantum7, Citrus limon6, Citrus paradisi6, Citrus sinensis14, Coffea arabica44, Eperua grandiflora46, Eucalyptus blakelyi4, Eucalyptus globulus5,7,56, Eucalyptus regnans87,88, Hedera helix7, Hypericum balearicum83, Laurus nobilis64, Ligustrum lucidum25,44, Limonium gibertii,79,83, Limonium magallufianum83, Macadamia integrifolia6, Metrosideros umbellate41, Nerium oleander7, Olea europea2,40,45,64,78,79, Pistacia lentiscus83, Quercus glauca25,44, Quercus ilex7,13,64,65,72, Quercus phillyraeoides25, Rhamnus alaternus52, Rhamnus ludovici-salvatoris52, Rhododendron macrophyllum54, Rosmarinus officinalis60, Weinmannia racemosa41; woody semi-deciduous angiosperms (n = 3): Phlomis italica83, Lavatera maritima83, Cistus albidus83; woody deciduous angiosperms (n = 37): Acer mono28,35, Acer mono var. marmoratum f. dissectum44, Acer circinatum54, Acer macrophyllum54, Acer palmatum35,44, Acer rufinerve35,44, Alnus japonica28,44, Alnus rhombifolia54, Alnus rubra54, Betula pendula36,50,51,63, Buddleja davidii74, Castanea sativa12,16,19, Corylus cornuta54, Fagus sylvatica12,86, Ficus carica14, Juglans nigra38, Juglans regia38, Liquidambar styraciflua48, Populus nigra55, Populus maximowiczii28,44, Populus deltoides13, Populus euramericana cv. Robusta18, Populus koreana x trichocarpa cv. Peace18,21, Populus tremeloides48, Populus trichocarpa x deltoides54, Prunus persica6, Quercus canariensis76, Quercus garryana54, Quercus petraea13, Quercus pubescens66, Quercus robur13,26,55, Quercus rubra5,7,54, Rhododendron occidentale54, Rosa rubiginosa21, Tilia cordata14,63, Vitis berlandieri × rupestris (Richter-110)79, Vitis vinifera34,42,47,52,68; herbaceous perennials (n = 13): Alocasia brisbanensis43, Beta maritima ssp. marcosii83, Beta maritima ssp. maritima83, Capsicum annuum27,33, Crepis triasii90, Digitalis minor var. minor84, Digitalis minor var. palauii84, Gossypium hirsutum19,61, Mentha spicata60, Origanum vulgare55, Phragmites australis31, Polygonum cuspidatum29, Solidago virgaurea63; herbaceous annual/biannual (n = 23): Arabidopsis thaliana79,80, Beta vulgaris7,71, Brassica oleracea89, Cucumis sativus7,48,79, Diplotaxis ibicensis83, Galinsoga ciliata55, Glycine max26,58,69, Helianthus annuus11,14,19,23,44, Ipomoea nil44, Lycopersicon esculentum44, Lysimachia minoricensis82, Nicotiana tabacum3,4,8,10,26,32,37,62,69,70,79, Nicotiana sylvestris52,73, Oryza sativa4,22,53, Phaseolus vulgaris4,41,44,48,52,81,86, Raphanus sativus4, Spinacia oleracea1,20,24,48,77, Triticum aestivum4, Triticum durum9, Triticum spp7,15, Vicia faba7,39, Vigna unguiculata14, Xanthium strumarium7,14; CAM plants (n = 1): Kalanchoë daigremontiana17,85. References: 1Evans & Terashima (1988); 2Bongi & Loreto (1989); 3Sharkey et al. (1991); 4von Caemmerer & Evans (1991); 5Harley et al. (1992); 6Lloyd et al. (1992); 7Loreto et al. (1992); 8Evans et al. (1994); 9Loreto et al. (1994); 10Price et al. (1994); 11Laisk & Sumberg (1994); 12Epron et al. (1995); 13Roupsard, Gross, Dreyer (1996); 14Laisk & Loreto (1996); 15Evans & Vellen (1996); 16Lauteri et al. (1997); 17Maxwell et al. (1997); 18Ridolfi & Dreyer (1997);19Brugnoli et al. (1998); 20Delfine et al. (1998); 21Genty et al. (1998); 22Scartazza et al. (1998); 23Eichelmann & Laisk (1999);24Delfine et al. (1999); 25Hanba, Miyazawa, Terashima (1999); 26Gillon & Yakir (2000); 27Delfine, Loreto, Alvino (2001); 28Hanba et al. (2001); 29Kogami et al. (2001); 30Miyazawa & Terashima (2001); 31Antonielli et al. (2002); 32Bernacchi et al. (2002); 33Delfine et al. (2002); 34Flexas et al. (2002); 35Hanba, Kogami & Terashima (2002); 36Laisk et al. (2002); 37Peisker & Apel (2001);38Piel et al. (2002); 39Terashima & Ono (2002); 40Centritto, Loreto & Chartzoulakis (2003); 41De Lucia, Whitehead & Clearwater (2003);42Düring (2003); 43Gorton et al. (2003); 44Hanba, Kogami & Terashima (2003); 45Loreto, Centritto & Chartzoulakis (2003); 46Pons & Welschen (2003); 47Sampol et al. (2003); 48Singsaas, Ort & Lucia (2003); 49Warren et al. (2003); 50Eichelmann et al. (2004a); 51Eichelmann et al. (2004b); 52Flexas et al. (2004); 53Hanba et al. (2004); 54Manter & Kerrigan (2004); 55Pons & Westbeek (2004); 56Warren (2004); 57Warren, Livingston & Turpin (2004); 58Bernacchi et al. (2005); 59Black et al. (2005); 60Delfine et al. (2005); 61Ennahli & Earl (2005); 62Galmés, Medrano & Flexas (2006);63Laisk et al. (2005); 64Niinemets et al. (2005); 65Peña-Rojas et al. (2005); 66Velikova et al. (2005); 67Ethier et al. (2006); 68Fila et al. (2006); 69Flexas et al. (2006a); 70Flexas et al. (2006b); 71Monti et al. (2006); 72Niinemets et al. (2006); 73Priault et al. (2006); 74Shi et al. (2006); 75Warren (2006b); 76Warren & Dreyer (2006); 77Yamori et al. (2006); 78Diaz-Espejo, Nicolás & Fernández (2007); 79Flexas et al. (2007a); 80Flexas et al. (2007b); 81Flowers et al. (2007); 82Galmés et al. (2007a); 83Galmés, Medrano, Flexas (2007b); 84Galmés, Medrano, Flexas (2007c); 85Griffiths et al. (2007); 86Warren et al. (2007); 87Warren (2008); 88Warren (unpublished); 89Flexas, Nogues & Cornic (unpublished results); 90Flexas et al. (unpublished results). gm, mesophyll conductance.

These pooled data suggest that gm is associated with leaf forms and plant functional groups, rather than reflecting evolutionary trends. For instance, in herbs, a big difference is observed between annuals/biannuals and perennials (i.e. functional groups) but not between monocots and dicots within each group (i.e. an evolutionary differentiation). In addition, CAM plants, which have evolved several times from C3 plants in relatively recent times, present values similar to those of the anciently evolved gymnosperms. Hence, it seems that fast-growing strategies like those of annual and biannual herbs are accompanied by high gm (and high photosynthesis rates) and vice versa. Remarkably, gm is quite low in evergreen trees and shrubs, and particularly in conifers, and this, in addition to low gs, may be the cause of their low photosynthesis rates and low photosynthetic nitrogen use efficiency (PNUE) (Manter & Kerrigan 2004; Warren & Adams 2004).

However, despite these general trends when considering data averaged per groups, significant variability can be found within a single group, genus or even species. For instance, within herbaceous annuals and biannuals, very large gm values (>1 mol CO2 m−2 s−1 bar−1) have been measured in fast-growing crops such as cotton (Brugnoli et al. 1998) and sunflower (Laisk & Loreto 1996), but a maximum gm as low as 0.08 mol CO2 m−2 s−1 bar−1 was found in the wild-extinct Mediterranean species Lysimachia minoricensis (Galmés et al. 2007a). The differences are even larger within woody deciduous angiosperms, with Betula pendula showing maximum gm > 1 mol CO2 m−2 s−1 bar−1 (Laisk et al. 2005) and Acer palmatum showing maximum gm < 0.06 mol CO2 m−2 s−1 bar−1 (Hanba et al. 2002, 2003). Although very few species have been analysed, there seems to be a large variability among semi-deciduous shrubs as well (Fig. 3). Within genus, some show a large variability, such as Citrus (0.02–0.42 mol CO2 m−2 s−1 bar−1), Populus (0.04–0.50 mol CO2 m−2 s−1 bar−1) or Quercus (0.07–0.30 mol CO2 m−2 s−1 bar−1), while in others such as Abies (0.02–0.13 mol CO2 m−2 s−1 bar−1), Acer (0.02–0.09 mol CO2 m−2 s−1 bar−1), Alnus (0.10–0.17 mol CO2 m−2 s−1 bar−1), Beta (0.18–0.34 mol CO2 m−2 s−1 bar−1), Eucalyptus (0.11–0.19 mol CO2 m−2 s−1 bar−1) or Pinus (0.04–0.17 mol CO2 m−2 s−1 bar−1), the variability is much smaller. Within species, a large variability is also found, particularly in cultivated species with a large number of cultivars such as bean (0.16–0.39 mol CO2 m−2 s−1 bar−1), cotton (0.3–1.8 mol CO2 m−2 s−1 bar−1), grapevines (0.07–0.30 mol CO2 m−2 s−1 bar−1), olives (0.08–0.35 mol CO2 m−2 s−1 bar−1), tobacco (0.09–0.50 mol CO2 m−2 s−1 bar−1) or wheat (0.20–0.64 mol CO2 m−2 s−1 bar−1).

Therefore, although general trends can be traced when comparing different functional groups, a large variability in gm is present within groups, genus and even species, suggesting that gm is a rapidly adapting trait likely being involved in the differences of photosynthetic efficiency found among different species and cultivars.


In addition to being largely variable among different species and cultivars, gm is currently known to change in response to a series of environmental variables. Early works have already suggested that gm decrease in response to water stress (Jones 1973), low nitrogen availability (Evans & Terashima 1988), salinity (Bongi & Loreto 1989) or high altitude (Vitousek, Field & Matson 1990). Now, it is known that gm not only decreases in the long term (days to weeks) in response to these environmental variables, but also to the exogenous application of abscisic acid (Flexas et al. 2006a) or poliethylenglycol (Warren et al. 2004), water logging (Black et al. 2005) or virus infections (Sampol et al. 2003). Shade leaves present lower gm than sun leaves (Hanba et al. 2002; Piel et al. 2002; Laisk et al. 2005; Warren et al. 2007). Moreover, gm increases when in vitro grown plants are acclimated to ex vitro conditions (Fila et al. 2006). These and other responses are summarized in Table 1. Long-term acclimation of gm to high CO2 and high O3 is less frequent (Singsaas et al. 2003; Eichelmann et al. 2004a; Bernacchi et al. 2005; Velikova et al. 2005; Flowers et al. 2007; Warren et al. 2007), although it occurs in some species and cultivars (Singsaas et al. 2003; Flowers et al. 2007). Acclimation to different growth temperatures is also complex. In spinach, plants grown at 30 and 15 °C attained similar maximum gm, but peak temperature was acclimated, occurring at 25 and 20 °C in high and low growth temperature plants, respectively (Yamori et al. 2006). On the other hand, in cabbage, both peak temperature and the maximum value of gm acclimate (Fig. 4; Flexas, Nogues & Cornic, unpublished results). In plants grown at only 5 °C, gm was severely depressed at any temperature, peaking at 18 °C, while in plants grown at 20 °C, gm values were normal and peaked at 24 °C. Leaf temperature also affected gm in olive leaves, with an optimum temperature at 29 °C (Diaz-Espejo et al. 2007). In this study, a seasonal evolution of diffusional limitations to photosynthesis under field conditions was suggested. On the other hand, in Eucalyptus regnans, no acclimation was observed for gm in plants grown at 15 or 30 °C (Warren 2008). Therefore, although gm responses to shade, nitrogen, water stress and salinity are quite generalized, further studies are needed to fully characterize its responses to carbon dioxide, ozone and temperature. This is necessary given the importance of these parameters for plant physiology in a climate change scenario. In addition, it would be interesting to analyse how gm is affected by nutrient deficiencies, other than nitrogen. In view of the possible role of carbonic anhydrase and aquaporins in regulating gm (see further), it would be particularly interesting to study the role of zinc, which is included in the molecular structure of carbonic anhydrases (Fabre et al. 2007), and phosphorus and calcium, which are involved in the regulation of aquaporin function (Kaldenhoff & Fischer 2006).

Table 1.  Compilation of published external and internal factors affecting gm
FactorTime scaleEffectReference(s)
  1. Internal factors have been divided into developmental, structural and metabolic.

  2. gm, mesophyll conductance; PAR, photosynthetically active radiation; ABA, abscisic acid; PEG, polyethylene glycol.

External factors
  Increased CO2 around leavesMinutes (up to 800 µmol CO2 mol−1 air)Slightly decreased gm in some species, not in othersHarley et al. (1992); Loreto et al. (1992)
Minutes (up to 1500 µmol CO2 mol−1 air)Large decrease of gmDüring (2003)
Minutes (up to 1500 µmol CO2 mol−1 air)Large decrease of gmFlexas et al. (2007a); Ethier & Pepin (personal communication); Niinemets (personal communication); Warren (personal communication)
Months (5 months)Unaffected gmEichelmann et al. (2004a)
Months (leaves fully developed at 550 µmol CO2 mol−1 air)Unaffected gmBernacchi et al. (2005)
Years (leaves fully developed at 560 µmol CO2 mol−1 air)Decreased gm in some species, not in othersSingsaas et al. (2003)
  Decreased CO2 around leavesMinutes (1 h)Increased gmCentritto et al. (2003)
Minutes (down to 50 µmol  CO2 mol−1 air)Large increases of gmFlexas et al. (2007a); Ethier & Pepin (personal communication); Niinemets (personal communication); Warren (personal communication)
Exogenous ABADays (3–7 days)Decreased gmFlexas et al. (2006a)
Exogenous HgCl2MinutesDecreased gmTerashima & Ono (2002)
Exogenous PEGDays (1–2 days)Decreased gmWarren et al. (2004)
Ex vitro acclimation of in vitro grown plantsMonths (2 months after transferred to ex vitro)Increased gmFila et al. (2006)
High altitudeConstitutive (plants grown at different altitudes)Decreased gmVitousek et al. (1990)
Constitutive (plants grown at different altitudes)Decreased gmKogami et al. (2001)
Constitutive (plants grown at different altitudes)Slightly increased gmShi et al. (2006)
High O3Days (3 days)Unaffected gmVelikova et al. (2005)
Months (5 months)Decreased gmEichelmann et al. (2004a)
Months (2 months) cultivars, not in othersDecreased gm in someFlowers et al. (2007)
Years (leaves fully developed at different O3 regimes)Unaffected gmWarren et al. (2007)
Light availabilityMinutesgm linearly and positively related to PARGorton et al. (2003)
Minutesgm linearly and positively related to PARFlexas et al. (2007a); Flexas et al. (unpublished results)
Constitutive (shade leaves)Decreased gmPiel et al. (2002)
Constitutive (shade leaves)Decreased gmHanba et al. (2002)
Constitutive (shade leaves)Decreased gmLaisk et al. (2005)
Constitutive (shade leaves)Decreased gmWarren et al. (2007)
Low N availabilityWeeks (50 days, plants growing at different nitrate concentrations)Decreased gmEvans & Terashima (1988)
Weeks (42 days, plants growing at different nitrate concentrations)Slightly decreased gmWarren (2004)
Constitutive (different wheat cultivars)Positive correlation between leaf N content and gmvon Caemmerer & Evans (1991)
SalinityDays (13 days)Decreased gmDelfine et al. (1998)
Weeks (90 days)Slightly decreased gmBongi & Loreto (1989)
Weeks (14–50 days)Decreased gmDelfine et al. (1999)
Weeks (40–45 days)Decreased gmCentritto et al. (2003); Loreto et al. (2003)
Temperature (T)Minutes to hoursIncreased gm from 10 to 35 °C, decreased at higher temperaturesBernacchi et al. (2002)
Minutes to hoursIncreased gm from 5 to 25 °C, then constant from 20 to 40 °CGorton et al. (2003)
Minutes to hoursIncreased gm from 10 to 20 °C, constant from 20 to 35 °CWarren & Dreyer (2006)
Minutes to hoursIncreased gm from 10 to 25 °C, constant or decreased (depending of acclimation) from 20 to 35 °CYamori et al. (2006)
Minutes to hoursIncreased gm from 10 to 35 °CWarren (2008)
Minutes to hoursIncreased gm from 10 to 18 or 25 °C, (depending of acclimation) and decreased thereafterFlexas et al. (unpublished results)
Hours (diurnal time course)Slight decrease of gm from 28 to 38 °CPons & Welschen (2003)
Weeks (plants grown at different temperatures for 3 weeks at 20 or 5 °C)Very low gm at any T with a peak at 18 °C (plants grown at low T) versus normal gm with a peak at 24 °C and decreased thereafter (plants grown at high T)Flexas et al. (unpublished results)
Months (plants grown at different temperatures for 2 to 4 months)Peak at 20 °C and decrease thereafter (plants grown at low T) versus peak at 25 °C and constant thereafter (plants grown at high T)Yamori et al. (2006)
Months (plants grown at 15 or 30 °C for 2 months)Increased gm from 10 to 35 °C in both plants grown at low and high TWarren (2008)
MonthsPeak at 29 °C and decrease thereafter (plants grown at high T)Diaz-Espejo et al. (2007)
Vapor pressure deficit (VPD)MinutesDecreased gm at increased VPDBongi & Loreto (1989)
MinutesUnaffected gmWarren (personal communication)
Virus infectionMonthsDecreased gmSampol et al. (2003)
WaterloggingWeeks (25 days)Decreased gmBlack et al. (2005)
Water stressMinutes (petiole cut in air)Decreased gmFlexas et al. (2006a)
Days (10 days)Decreased gmRoupsard et al. (1996)
Days (10 days)Decreased gmRidolfi & Dreyer (1997)
Days (5 days)Decreased gmFlexas et al. (2004)
Days (9 days)Decreased gmEnnahli & Earl (2005)
Days (4–13 days)Decreased gmGalmés et al. (2007b)
Days (5 days)Decreased gmGalmés et al. (2007c)
Weeks (15 days)Decreased gmBrugnoli et al. (1998)
Weeks (20–70 days)Decreased gmScartazza et al. (1998)
Weeks (14–42 days)Unaffected gmDelfine et al. (2001, 2002)
Weeks (66 days)Decreased gmDelfine et al. (2001, 2002)
Weeks (15–60 days)Decreased gmFlexas et al. (2002)
Weeks (60 days)Decreased gmDelfine et al. (2005)
Weeks (30–80 days)Decreased gmGrassi & Magnani (2005)
Weeks (20–50 days)Decreased gmPeña-Rojas et al. (2005)
Weeks (14 days)Decreased gmBlack et al. (2005)
Weeks (21 days)Decreased gmGalmés et al. (2007a)
Weeks (37 days)Unaffected gmMonti et al. (2006)
Weeks (37–70 days)Decreased gmMonti et al. (2006)
Weeks (25–45 days)Decreased gmRipley et al. (2007)
Months (leaves totally developed under water stress)Decreased gmGalmés et al. (2006)
Internal factors
 Developmental factors
  Leaf ageingDays (12 days)Decreased gmFlexas et al. (2007b)
Weeks (25 days)Decreased gmLoreto et al. (1994)
Weeks (20–70 days)Decreased gmScartazza et al. (1998)
Weeks (14–50 days)Decreased gmDelfine et al. (1999)
Weeks (25 days)Decreased gmGrassi & Magnani (2005)
Weeks (30–40 days)Decreased gmBernacchi et al. (2005)
Years (1 to 6 years in evergreen trees)Decreased gmNiinemets et al. (2005, 2006)
Years (1 to 3 years in evergreen trees)Slightly decreased gmWarren (2006b)
Years (1 to 3 years in evergreen trees)Decreased gmEthier et al. (2006)
 Leaf developmentDays to weeks (10–30 days)Increased gmMiyazawa & Terashima (2001)
Days (15 days)Increased gmEichelmann et al. (2004b)
Structural factors
 Chloroplast movementsMinutes (transfer from low to high light)Unaffected gmGorton et al. (2003)
Minutes (different light treatments)Affected gmTholen et al. (2007)
 Chloroplast rearrangementsConstitutive (phytochrome mutants)Decreased gm in some experiments, not in othersSharkey et al. (1991)
 Chloroplast surface exposed to intercellular air spacesDays to weeks (10–30 days, changes due to leaf development)Positively correlated with gmMiyazawa & Terashima (2001)
Constitutive (wheat)Positively correlated with gmvon Caemmerer & Evans (1991)
Constitutive (tobacco rubisco mutants)Positively correlated with gmEvans et al. (1994)
Constitutive (different woody species)Positively correlated with gmLloyd et al. (1992)/Syvertsen et al. (1995)
Constitutive (different wheat cultivars)No correlation with gmEvans & Vellen (1996)
Constitutive (different species) plus age-inducedPositively correlated with gmHanba et al. (2001)
Constitutive (plants grown at different altitudes)No correlation with gmKogami et al. (2001)
Constitutive (leaves grown at different light intensities)No correlation with gmHanba et al. (2002)
Constitutive (transgenic plants HvPIP2;1)No correlation with gmHanba et al. (2004)
 Leaf porosityDays to weeks (10–30 days, changes due to leaf development)Positively correlated with gmMiyazawa & Terashima (2001)
Constitutive (different species)Positively correlated with gmLoreto et al. (1992)
Constitutive (different tobacco mutants)Negatively correlated with gmEvans et al. (1994)
Constitutive (different evergreen trees)Negatively correlated with gmHanba et al. (1999)
Constitutive (different species) plus age-inducedNo correlation with gmHanba et al. (2001)
Constitutive (plants grown at different altitudes)Negatively correlated with gmKogami et al. (2001)
Constitutive (leaves grown at different light intensities)Negatively correlated with gm in some species (no correlation in others)Hanba et al. (2002)
Constitutive (transgenic plants HvPIP2;1)Negatively correlated with gmHanba et al. 2004)
Developmental (resprout leaves as compared to undisturbed plants)Positively correlated with gmPeña-Rojas et al. 2005)
 Mesophyll surface exposed to intercellular air spacesDays to weeks (10–30 days, changes due to leaf development)Positively correlated with gmMiyazawa & Terashima (2001)
Constitutive (different woody species)Positively correlated with gmLloyd et al. (1992); Syvertsen et al. (1995)
Constitutive (different evergreen trees)Positively correlated with gmHanba et al. (1999)
Constitutive (different species) plus age-inducedNo correlation with gmHanba et al. (2001)
Constitutive (plants grown at different altitudes)Positively correlated with gmKogami et al. (2001)
Constitutive (transgenic plants HvPIP2;1)Negatively correlated with gmHanba et al. (2004)
Metabolic factors
  Inhibition of tobacco aquaporin NtAQP1Constitutive (antisense plants)Decreased gmFlexas et al. (2006b)
  Over-expression of barley aquaporin HvPIP2 in riceConstitutive (transgenic plants)Increased gmHanba et al. (2004)
  Over-expression of tobacco NtAQP1Days to weeks (gene inducible with tetracycline)Increased gmFlexas et al. (2006b)
Carbonic anhydrase activityConstitutive (CA mutants)Positively correlated with small (25%) variations of gmPrice et al. (1994)
Constitutive (three different species)Positively correlated with the chloroplast fraction of gmGillon & Yakir (2000)
Mitochondrial mutations
 Lack of Complex IConstitutive (CMSI mutation)Decreased gmPriault et al. (2006)
 Impaired Complex IConstitutive (MSC16 mutation)Decreased gmJuszczuk et al. (2007)
Figure 4.

(a) Simultaneous response of stomatal (triangles) and mesophyll (circles) conductance to cutting a Glycine max leaf through the petiole in air. Data redrawn from Flexas et al. (2006a). (b) Responses of gm to leaf temperature in Brassica oleracea plants grown either at 5 °C (empty circles) or 20 °C (filled circles). Temperature was decreased stepwise from 30 to 10 °C at 30 min intervals. Measurements were carried out at constant light intensity and vapour pressure deficit (VPD). Data from Flexas, Nogués & Cornic (unpublished results). (c) The response of gm to varying Ci in six species: Olea europaea, Cucumis sativus, Nicotiana tabaccum, Vitis berlandieri × rupestris, Limonium gibertii and Arabidopsis thaliana. Ci was changed by varying Ca stepwise at 3 min intervals. Measurements were carried out at constant light intensity, temperature and VPD. Data from Flexas et al. (2007a). (d) The response of gm to varying light intensity in N. tabaccum. Redrawn from Flexas et al. (2007a). gm, mesophyll conductance; gs, stomatal conductance; PPFD, photosynthetic photon flux density.

Moreover, many studies have shown that leaf development and ageing strongly influence gm (Table 1). Hence, during leaf development from unfolding to maturation, gm increases in parallel with leaf photosynthetic capacity (Miyazawa & Terashima 2001; Eichelmann et al. 2004b). Contrarily, leaf ageing results in decreased gm, regardless of whether ageing takes place in a few days, as in Arabidopsis (Flexas et al. 2007b), in weeks, as in some herbs (Loreto et al. 1994; Scartazza et al. 1998; Delfine et al. 1999; Bernacchi et al. 2005) and deciduous trees (Grassi & Magnani 2005), or along years, as in sclerophyll evergreen shrubs and trees (Niinemets et al. 2005, 2006; Ethier et al. 2006; Warren 2006b). Indeed, reduced gm seems to be the most important cause of early photosynthetic reductions in ageing leaves in herbs (Loreto et al. 1994; Delfine et al. 1999; Flexas et al. 2007b) and woody evergreen plants (Niinemets et al. 2005, 2006; Ethier et al. 2006), although not in deciduous trees (Grassi & Magnani 2005).

In addition to long-term and developmental responses, gm seems also able to respond in the short time (minutes) to several environmental variables. Figure 4 shows examples of these rapid responses, such as rapid leaf desiccation, changes in leaf temperature, changes in CO2 concentration and changes in light intensity. Figure 4a illustrates the speed of gm down-regulation. In soybean leaves, upon cutting the leaf petiole in air, there is a parallel decline of both gs and gm of about 30% after only 10 min. In less than 1 h, both conductances were almost zero. In cabbage, inducing changes in leaf temperature also results in large responses of gm within 20–30 min, in plants acclimated to both low and high temperatures (Fig. 4b). Similar responses have been observed in tobacco (Bernacchi et al. 2002), Alocasia brisbanensis (Gorton et al. 2003), spinach (Yamori et al. 2006), Quercus canariensis (Warren & Dreyer 2006) and E. regnans (Warren 2008). Rapid changes in CO2 concentration around leaves, such as those typically done when performing AN-Ci curves (i.e. within a few minutes), also result in strong responses of gm (Fig. 4c). Remarkably, when scaled as percent of maximum values, six different species belonging to different functional groups show almost identical response to Ci (Fig. 4c), reinforcing the generality of this relationship. Similar responses have been recently observed in other species (Ethier and Pepin, Niinemets, Warren, personal communication). Studies on the kinetics of gm and gs recovery upon returning from high to ambient CO2 suggest that gm is actually faster than gs (Flexas et al. 2007a). It was also suggested that light intensity during measurements also affects the magnitude of gm (Fig. 4d). Actually, a linear positive relationship between gliq and light intensity could be obtained from a re-analysis of the data by Gorton et al. (2003).

Therefore, it seems that gm and stomata respond to all the same environmental variables, and in a similar manner (i.e. decreasing in response to water loss, decreased temperature, increased CO2 and decreased light). An exception could be vapour pressure deficit (VPD). Although Bongi & Loreto (1989) showed a strong reduction of gm in olive in response to increased VPD, similar to the stomatal response, recent studies by Warren in Eucalyptus, beans and tomato showed that gs decreased upon increasing VPD, while gm does not (Warren, personal communication). Although more data are needed on the response of gm to light and VPD to clarify these responses, it may be concluded that gm and gs are tightly co-regulated under most circumstances, resulting in adjustments in CO2 availability in the chloroplasts in response to environmental changes.

In summary, a large body of evidence has accumulated in recent years to conclude that gm is not only finite, but it also acclimates and responds both in the long (days, weeks) and short (minutes, hours) terms to many environmental variables, including light, temperature, water and CO2. These changes may be important in regulating photosynthesis in response to the environment.


Despite substantial evidence for large variability of gm among species and genotypes, and for both long-term and rapid responses to environmental factors, the mechanistic basis of these variations remains unclear. Early literature has assumed that simple diffusion through cellular membranes (Colman & Espie 1985) and/or leaf structural properties (Lloyd et al. 1992; Evans et al. 1994; Syvertsen et al. 1995) are causing most variations in gm. In fact, gm has been shown to correlate with some leaf structural properties (see Table 1), such as phytochrome-related mutation-induced chloroplast rearrangements (Sharkey et al. 1991), mesophyll (Smes) and/or chloroplast (Sc) surfaces directly exposed to intercellular air spaces (Evans & Loreto 2000), and chloroplast development (Tholen 2005) and movements (Tholen et al. 2007). However, other studies have shown no relationship between gm and these leaf traits (Evans & Vellen 1996; Hanba et al. 2001, 2002, 2004; Kogami et al. 2001; Gorton et al. 2003). Other leaf traits, such as leaf porosity, are in some studies positively correlated with gm (Loreto et al. 1992; Miyazawa & Terashima 2001; Peña-Rojas et al. 2005), while in some others are negatively correlated (Evans et al. 1994; Hanba et al. 1999; Kogami et al. 2001; Hanba et al. 2002, 2004). Moreover, Terashima et al. (2005) have shown that the slope of the relationship between gm and Smes decreases from annual herbs to evergreen trees, with deciduous trees showing intermediate values. Considering only literature data from plants in the absence of any stress, the relationship between gm and leaf mass per area (LMA), the simplest indicator of leaf structure, is asymptotic (Fig. 5), that is, leaf structure sets the limit for maximum gm, but not its actual value. Hence, while leaf structure strongly limits gm in evergreen species with large LMA, gm presents a large range of variation in mesophyte species with low LMA. Moreover, while structural properties could certainly be involved in adaptive and acclimation responses, they could not account for the rapid variations observed in response to varying environmental conditions.

Figure 5.

The relationship between mesophyll conductance (gm) and leaf mass per area (LMA) in different species, in the absence of stress. Data from high-altitude plants are indicated by filled upward triangles, old leaves by filled downward triangles and shade leaves by filled squares. Other leaves are indicated by empty circles. Data have been compiled from the following references: Vitousek et al. (1990), Syvertsen et al. (1995), Lauteri et al. (1997), Hanba et al. (1999), Kogami et al. (2001), Piel et al. (2002), Flexas et al. (2002, 2004, 2006b), De Lucia et al. (2003), Warren et al. (2003), Grassi & Magnani (2005), Niinemets et al. (2005, 2006), Priault et al. (2006), Shi et al. (2006), Galmés et al. (2007b).

It is important to stress that leaf structure may affect mostly gias and gw. Early studies comparing CO2 leaf diffusion in air and helox suggested that gias was low and limited photosynthesis, particularly in hypostomatous leaves (Parkhurst & Mott 1990). Moreover, in early studies, gw was sometimes used as a synonymous of gm, because it was assumed that cell wall exerted most of mesophyll resistance to CO2 transfer (Bongi & Loreto 1989; von Caemmerer & Evans 1991; Lloyd et al. 1992). Therefore, leaf structure was thought to be the main determinant of gm. However, the CO2 diffusion coefficient in the liquid phase is much slower than in air, suggesting that gliq must be low (Nobel 1983). Actually, a re-evaluation of gias using the comparison of diffusivities in air and helox, revealed that gliq was much smaller than gias, and therefore the most limiting factor for photosynthesis (Genty et al. 1998; Piel 2002). Using a photoacoustic technique, Gorton et al. (2003) also concluded that low gliq was the most limiting factor for CO2 diffusion in the mesophyll. Combining 13C and 18O discrimination by leaves, Gillon & Yakir (2000) were able to partition gm into two components, gw and a cellular component that was termed chloroplast conductance (gchl) instead of gliq because, as chloroplasts are usually tightly coupled to cell membranes facing intercellular air spaces (Evans & von Caemmerer 1996), it was assumed that CO2 would not have to cross the cytosol, entering the chloroplasts directly after crossing the cell wall and plasma and chloroplast membranes. These authors showed that gw was lower than gchl in thick leaves of oaks, but gchl was lower than gw in the mesophytic leaves of soybean and tobacco. In contrast, Piel (2002) showed that even in the sclerophyll oak Quercus ilex rliq (the inverse of gliq) contributed to 70% of the total internal resistance, while rias (the inverse of gias) contributed the remaining 30%. Therefore, gliq (or gchl) seems the most limiting gm component, sure in mesophytic species but probably also in sclerophylls.

It is likely that a metabolic process might be involved in gliq (or gchl) variations. Based on a temperature response coefficient (Q10) of approximately 2.2 for gm in tobacco leaves, Bernacchi et al. (2002) speculated that there must be an enzymatic or protein-facilitated diffusion control of gm. The most likely candidates for the most dynamic gm changes could be carbonic anhydrase and aquaporins.

Some authors have suggested that carbonic anhydrase activity is closely associated with gm in C3 plants (Makino et al. 1992; Sasaki, Samejima & Ishii 1996). However, modification of carbonic anhydrase activity in transgenic plants revealed some (about 25%) or no change in gm and photosynthesis (Price et al. 1994; Williams, Flanagan & Coleman 1996a), and regression analysis showed only a modest correlation between carbonic anhydrase activity and photosynthesis in different families of Tectona grandis (Tiwari et al. 2006). An explanation was provided by Gillon & Yakir (2000), who showed that the relative contribution of carbonic anhydrase to the overall gm is species dependent. They suggested that carbonic anhydrase-mediated CO2 diffusion may be more important when gm is low because of structural properties of the leaves, as is the case for woody species, where cell wall conductance is much lower than chloroplast conductance. Moreover, it has recently been shown in Arabidopsis that there is a much larger amount of genes encoding for different carbonic anhydrases than previously thought (Fabre et al. 2007). These authors have suggested that several different carbonic anhydrases, located in the plasma membrane, cytosol and chloroplasts, could contribute to increase gm. The possible role of different carbonic anhydrases on gm is currently being studied, although preliminary results have shown little evidence for the involvement of the most abundant forms in gm (Genty, personal communication).

Regarding a possible role for aquaporins, the first indirect evidence was provided by Terashima & Ono (2002), who impaired mesophyll conductance to CO2 by HgCl2 (a non-specific inhibitor of some aquaporins). Uehlein et al. (2003) demonstrated that tobacco aquaporin NtAQP1 facilitates trans-membrane CO2 transport by expression in Xenopus oocytes. More recently, substantial evidence has been compiled, clearly demonstrating a role of some specific aquaporins in the regulation of gm (Hanba et al. 2004; Flexas et al. 2006b). Hanba et al. (2004) found that transgenic rice plants over-expressing the barley aquaporin HvPIP2;1 not only presented higher gm, but also anatomical (Sm, Sc, mesophyll porosity, stomatal density, stomatal size) and physiological (concentration of Rubisco) differences. Therefore, it was not clear whether the aquaporin had a direct effect on gm, or whether it was an indirect effect caused by anatomical and physiological differences induced by plant transformation. However, such anatomical and physiological differences were not induced when the expression of native NtAQP1 was blocked or over-expressed in tobacco, while still important differences in gm were associated with aquaporin levels (Flexas et al. 2006b). These results, together with the fact that NtAQP1 is permeable to CO2, suggest a direct involvement of aquaporins in gm. Flexas et al. (2007a) has recently suggested that NtAQP1 could be involved in the rapid (minutes) response of gm to changes in CO2 concentration. Although the regulation of aquaporin activity in the short term is not fully understood, several mechanisms have been proposed, including direct phosphorylation of aquaporins (Kjellbom et al. 1999), an osmotically driven cohesion/tension mechanism (Ye, Wiera & Steudle 2004), pH-dependent gating of aquaporins (Tournaire-Roux et al. 2003) and transcriptional regulation and protein stability (Eckert et al. 1999). CO2-transporting aquaporins are typically located in plasma membranes (Kaldenhoff & Fischer 2006), although recently, NtAQP1 has been shown to be present in the inner chloroplast membrane as well (Kaldenhoff, personal communication). Therefore, as carbonic anhydrases, aquaporins could be involved in gliq by acting at several points along the CO2 path.

In addition to carbonic anhydrases and aquaporins, other leaf internal factors could be involved in gm. There have been already mentioned the cases of tobacco mutants expressing excess phytochrome (Sharkey et al. 1991) and Arabidopsis mutants unresponsive to ethylene (Tholen 2005; Tholen et al. 2007), although in these cases, altered gm seems to be a consequence of chloroplast rearrangements or movements induced by the mutation and not a direct effect of the mutation on gm. More intriguing is the case of mitochondrial mutants, such as the cytoplasmic male sterile II (CMSII) mutant of Nicotiana sylvestris lacking complex I of the mitochondrial electron transport chain (Priault et al. 2006), or the mosaic mutant of cucumber (MSC16) also impaired at the level of complex I (Juszczuk et al. 2007). Both mutants present lower photosynthetic capacity as compared with wild type, and in both cases, this decrease is due to a reduced gm (Priault et al. 2006) or reduced gs plus gm (Juszczuk et al. 2007). Clearly, the physiological link between mitochondrial electron transport chain and down-regulation of gm needs to be elucidated.

Summarizing, despite substantial evidence for large variability of gm, the mechanistic basis of these variations remains unclear. While structural properties could be involved in adaptive and acclimation responses, the most likely candidates for the most dynamic gm changes would be carbonic anhydrase and aquaporins, although other components cannot be discarded at present.


From a global perspective, the fact that leaf thickness and density (i.e. LMA) set a limit for the maximum gm (Fig. 5) contributes to establish the so-called ‘worldwide leaf economics spectrum’ (Wright et al. 2004) and, particularly, the negative relationship between AN and/or PNUE and LMA (Hikosaka et al. 1998; Hikosaka 2004; Pons & Westbeek 2004; Warren & Adams 2006). In other words, interspecific differences in gm define ecological strategies along a gradient from fast-growing, high PNUE species with a low persistence under stress to slow-growing, low PNUE species with a high persistence under stress. Differences in gm among varieties or provenances of a given species have also been described and related to differences in photosynthetic efficiency (Evans & Vellen 1996; Lauteri et al. 1997; Patakas, Kofidis & Bosabalidis 2003). Evolutionary studies have attempted to describe whether adapting stomatal conductance and/or leaf biochemical capacity is effective in the generation of photosythetically efficient species (Geber & Dawson 1990) or crop cultivars (Koç, Barutçular & Genç 2003). Further studies would be necessary to establish how adapting gm can also contribute to the evolution of photosynthetically efficient plants.

A particular aspect of photosynthetic efficiency, of enormous importance in semi-arid and arid environments, is leaf WUE. At least theoretically, increasing AN by means of increasing gs has a cost in terms of transpiration, leading to decreased WUE. On the contrary, increasing AN by means of increasing gm may result in increased WUE (Warren & Adams 2006; Aranda et al. 2007). Evans & Vellen (1996) showed that high WUE was related to a high gm in some wheat cultivars, but not in others. Similarly, Lauteri et al. (1997) showed that provenances of Castanea sativa from low rainfall areas had a higher WUE and a higher gm than provenances from high rainfall areas. In plants from Bursa (mean annual precipitation = 700 mm), the ratio gm/gs was 0.98, decreasing to 0.88 and 0.49 in plants from Hopa and Girseum, respectively (mean annual precipitation = 1125 mm). A similar correspondence between WUE and gm/gs was observed in provenances of the annual herb Crepis tirasii, an ancient endemism from the Balearic Islands with a very fragmentary distribution in isolated populations (Flexas, Galmés, Mayol and Riba, unpublished results). When growing in a common environment, plants from six different populations from sites with precipitation ranging 400–600 mm year−1 showed a very similar leaf intrinsic WUE (AN/gs = 26–30 µmol CO2 mol−1 H2O). These plants presented gm/gs ratios between 0.3 and 0.4 mol mol−1 (Fig. 6). However, a seventh separated population from Cabrera, a small islet with annual precipitation lower than 300 mm year−1, showed a 25% larger WUE (AN/gs = 40 µmol CO2 mol−1 H2O), in correspondence with increased gm/gs (0.5 mol mol−1).

Figure 6.

The relationship between net photosynthesis (AN) and stomatal conductance (gs) in plants of Crepis triasii from seven different populations from the Balearic Islands grown in a common environment in Cambrils. Six of the populations (filled circles) came from sites with annual precipitation ranging from 400 to 600 mm year−1, while another population (empty circle) came from Cabrera, a small islet with an annual precipitation lower than 600 mm year−1. The values of AN/gs and gm/gs are given. Data are average ± SE of five to eight replicates per population. gm, mesophyll conductance.

In summary, gm adjustments could play a role in defining the photosynthetic strategies of plants, contributing to define their ecological niches. In addition, increasing the ratio gm/gs enhances photosynthesis and WUE in plants evolved under arid environments. While further studies including more species and larger numbers of cultivars are needed, these results suggest that gm could be a good target for the improvement of crop WUE through biotechnology.


The current knowledge that gm is finite and variable, and that Ci differs from Cc in a manner that is not easily predictable strengthens the need to incorporate a term that considers gm in current photosynthesis models, such as that by Farquhar et al. (1980), as already suggested (Bernacchi et al. 2002; Flexas et al. 2002; Long & Bernacchi 2003; Ethier & Livingston 2004; Manter & Kerrigan 2004; Sharkey et al. 2007). In general, ignoring the difference between Cc and Ci caused by gm would result in an underestimation of the maximum velocity of carboxylation (Vc,max) and, to a lesser extent, underestimation of the maximum capacity for electron transport (Jmax). This effect is very important particularly when gm is largely reduced, such as under severe water stress. Under these conditions, using Ci instead of Cc can lead to errors as large as ca. 100% in the estimation of Vc,max, as illustrated by Flexas et al. (2006c, 2007a).

The importance of AN-Ci curves and photosynthesis models relies on the fact that they are commonly used to develop prediction models of CO2 assimilation for crops (Diaz-Espejo et al. 2006) and natural vegetation (Xu & Baldocchi 2003), to assess the influence of stresses on the photosynthetic capacity of plants (Centritto et al. 2003; Loreto et al. 2003) and to help predict the effects of climate change on photosynthesis (Rogers, Ellsworth & Humphries 2001). Moreover, these measurements and models are the basis for scaling up from leaf to whole plant and/or ecosystem carbon assimilation models (Harley & Baldocchi 1995). While gm has started being taken into account in leaf-level photosynthesis parameterization studies (Ethier & Livingston 2004; Manter & Kerrigan 2004; Flexas et al. 2006c, 2007a,b; Diaz-Espejo et al. 2007; Sharkey et al. 2007), to the best of our knowledge, it was considered in only two studies concerning whole-plant modelling (Williams et al. 1996b; Le Roux et al. 2001). In the first of these studies, gm was considered as a constant, while in the second, it was derived from a previously established linear relationship between gm and Vc,max. However, this may not represent a sufficiently accurate approach, and studies have begun to address how gm at different canopy levels responds to the interaction of leaf age and illumination history (Niinemets et al. 2006) or water stress and canopy microclimate (Diaz-Espejo et al. 2007). Clearly, more studies are needed in this direction, as well as modelling efforts to include a finite and variable gm in whole-plant photosynthesis models.

In addition to photosynthesis models, carbon isotope discrimination models (Farquhar, O'Leary & Berry 1982; Evans et al. 1986) may also be affected by gm. While Evans et al. (1986) already included a term for gm in their isotope discrimination model, in practice, this is often neglected. It is usually assumed that most of the leaf discrimination against 13C is due to discrimination by Rubisco and phosphoenolpyruvate (PEP) carboxylase, with another minor fractionation caused by CO2 diffusion in air through the stomata. As a result, 13C discrimination (Δ) is proportional to the Ci/Ca ratio, and δ13C in leaf dry matter is interpreted as resulting from the ‘mean leaf-life’Ci/Ca ratio, which can be related to intrinsic WUE (AN/gs). Therefore, δ13C in leaf dry matter has been used as a long-term estimation to compare WUE between species or genotypes. However, differences in δ13C may correlate more precisely with Cc/Ca than with Ci/Ca (Le Roux et al. 2001). Theoretical estimations reveal that differences in leaf dry matter δ13C as large as 2–4‰ can be accounted by differences in gm, with no difference in WUE (Le Roux et al. 2001; Warren & Adams 2006). Experimental data by Hanba et al. (2003) confirmed these estimations. These authors showed that highland and lowland Polygonum plants presented identical AN/gs but differed by >3‰ in leaf dry matter δ13C because of large differences in gm. Similar results were observed by Flexas et al. (2006b) in transgenic tobacco plants specifically affected in gm through modification of aquaporin NtAQP1 expression (Fig. 7). Substantial differences (1.5‰) were found in leaf δ13C between different lines with only small changes in their Ci/Ca ratio, which weakens the correlation between δ13C and WUE (Fig. 7a), while δ13C was strongly correlated with Cc/Ca (Fig. 7b) and gm (Fig. 7c).

Figure 7.

The relationship between δ13C in leaf dry matter and the instantaneous ratios Ci/Ca (a) and Cc/Ci (b), and gm (c) in transgenic tobacco differing in the expression of aquaporin NtAQP1. Symbols indicate wild-type (CAS, empty circles or CO, empty triangles), antisense (AS, filled circles) and over-expressing (O, filled triangles) plants. Values are means ± SE of six replicates per genotype. Redrawn from Flexas et al. (2006b). gm, mesophyll conductance.

Therefore, the existing evidence supports the idea that discrimination during CO2 diffusion within the leaf should not be neglected when using carbon isotope discrimination models to assess plant WUE.


Substantial evidence has accumulated in the past two decades showing that mesophyll conductance to CO2 is finite and highly variable, and that it reduces Cc below Ci, therefore limiting photosynthesis. gm has been determined in more than a hundred species, and under several environmental conditions. The present review shows that gm differs among plant functional groups, and that it acclimates and responds both in the long (days, weeks) and short (minutes, hours) terms to changes in environmental variables such as light, temperature, CO2, nitrogen and water availability, among others, as well as with leaf development and ageing.

Concerning the factors that determine gm, it has been shown that leaf structural traits determine gm only partially, while an important part of its regulation seems to have a metabolic origin, possibly related to aquaporins and/or carbonic anhydrase, and perhaps other physiological traits. From the ecological viewpoint, it has been stressed that gm can help in understanding the ecological and evolutionary aspects of plants. On the other hand, from the agronomical viewpoint, gm could help to improve WUE more efficiently than to modify gs. Finally, it has been discussed how ignoring gm would result in errors in photosynthesis parameterization, modelling and scaling up from leaf to whole plant, as well as lead to misleading interpretations of carbon isotope discrimination data.

Some research priorities for the near future must be highlighted. Among these are

  • 1To further analyse how gm responds to environmental factors, particularly to temperature, VPD and deficiencies in different nutrients. We suggest that studying the responses to deficiencies of phosphorous, zinc and calcium may be of particular interest given the involvement of these elements in the function of aquaporins and/or carbonic anhydrases.
  • 2To fully elucidate the physiological and molecular mechanisms involved in gm regulation, and especially how these respond to environmental changes. To achieve this goal, different approaches may be used: firstly, to analyse the response of gm to different plant hormones, in order to understand the patterns of its regulation; secondly, to perform ecophysiological studies where variations of gm are determined together with variations in amounts and/or activities of aquaporins and carbonic anhydrase; finally, to increase the availability of transgenic plants with alterations in carbonic anhydrases, aquaporins and other factors, to characterize the metabolic components directly involved in the regulation of gm.
  • 3To increase our knowledge of how changes in gm lead to the evolution of genotypes with increased photosynthetic nitrogen and/or WUE. It would be especially desirable to undertake broad comparisons of different natural plant populations adapted to different environments and of different cultivars of crop species with different adaptations.
  • 4To improve photosynthesis and isotope discrimination models to consider gm in an easy but accurate manner, in order to increase their accuracy and to improve predictions of canopy photosynthesis, plant productivity and plant responses to climate change.


Our studies on mesophyll conductance to CO2 were partly granted by projects BFI2002-00772 and BFU2005-03102/BFI (Plan Nacional, Spain). M.R.-C. and A.D.-E. were beneficiaries of the ‘Programa Ramón y Cajal’ (M.E.C.), and J.G. was granted a postdoctoral fellowship (M.E.C.). We acknowledge helpful discussions about this subject with Drs Tom Sharkey, Francesco Loreto, Joe Berry, Gabriel Cornic and Ichiro Terashima. We also acknowledge access to unpublished results and manuscripts, data bases and clarification about published issues to Drs Gilbert Ethier, John Evans, Bernard Genty, Ralf Kaldenhoff, Agu Laisk, Ülo Niinemets, Danny Tholen and Charles Warren.