Intramolecular, compound-specific, and bulk carbon isotope patterns in C3 and C4 plants: a review and synthesis


  • Erik A. Hobbie,

    Corresponding author
    1. Max-Planck-Institute for Biogeochemistry, Jena, Germany; Present address: Complex Systems Research Center, Morse Hall, University of New Hampshire, Durham, New Hampshire 03824-3525, USA;
      Author for correspondence: Erik A. Hobbie Tel: +1 603 862 3581 Fax: +1 603 862 0188 Email:
    Search for more papers by this author
  • Roland A. Werner

    1. Present address: Institut für Pflanzenwissenschaften ETH Zentrum, LFW C48.1, Universitätsstrasse Z, 8092 Zürich, Switzerland
    Search for more papers by this author

Author for correspondence: Erik A. Hobbie Tel: +1 603 862 3581 Fax: +1 603 862 0188 Email:



  • Summary 371

  • I. Introduction 372
  • II. Methods and terminology 373
  • III. Results 373
  • IV. Discussion 376
  • V. Conclusions 382
  • Acknowledgements 382

  • References 382


Studies using carbon isotope differences between C3 and C4 photosynthesis to calculate terrestrial productivity or soil carbon turnover assume that intramolecular isotopic patterns and isotopic shifts between specific plant components are similar in C3 and C4 plants. To test these assumptions, we calculated isotopic differences in studies measuring components from C3 or C4 photosynthesis. Relative to source sugars in fermentation, C3-derived ethanol had less 13C and C3-derived CO2 had more 13C than C4-derived ethanol and CO2. Both results agreed with intramolecular isotopic signatures in C3 and C4 glucose. Isotopic shifts between plant compounds (e.g. lignin and cellulose) or tissues (e.g. leaves and roots) also differed in C3 and C4 plants. Woody C3 plants allocated more carbon to 13C-depleted compounds such as lignin or lipids than herbaceous C3 or C4 plants. This allocation influenced 13C patterns among compounds and tissues. Photorespiration and isotopic fractionation at metabolic branch points, coupled to different allocation patterns during metabolism for C3 vs C4 plants, probably influence position-specific and compound-specific isotopic differences. Differing 13C content of mobile and immobile compounds (e.g. sugars vs lignin) may then create isotopic differences among plant pools and along transport pathways. We conclude that a few basic mechanisms can explain intramolecular, compound-specific and bulk isotopic differences between C3 and C4 plants. Understanding these mechanisms will improve our ability to link bulk and compound-specific isotopic patterns to metabolic pathways in C3 and C4 plants.

I. Introduction

Perhaps the most important physiological division in higher plants is between those using the C3 vs the C4 pathway of carbon fixation during photosynthesis, with the C4 pathway a recent (≈ 20 million yr) adaptation to conserving water and reducing photorespiration in dry climates. Because of different opportunities for isotopic fractionation in the initial fixation of CO2 in the two pathways, the 13C : 12C ratios also differ, with C3 plants depleted in 13C relative to C4 plants. A rich literature has developed over the past 30 yr explaining isotopic patterns in C3 and C4 plants at the bulk level using analytical equations (Vogel, 1993), whereas an equally rich literature has used isotopic differences between C3 and C4 plants to examine the relative contributions of C3- or C4-derived plant matter to modern diets and paleodiets (Cerling & Harris, 1999; Cerling et al., 1999), to soil organic matter formation (Balesdent & Mariotti, 1996), or to the global carbon cycle (Lloyd & Farquhar, 1994). However, it is generally not appreciated that the two photosynthetic pathways also often differ in shifts in 13C content between specific plant tissues, between specific compounds, or at the intramolecular level. Although such shifts in 13C content may accordingly cause errors in studies assuming similar patterns of isotopic distribution in the C3 and C4 pathway, they may also provide important clues to mechanisms of formation and transport of carbon within plants. In this paper we examine evidence from the literature for differences in 13C shifts between C3 and C4 systems at tissue-, compound- and position-specific levels, and discuss possible mechanisms creating such differences. A deeper understanding of how such differences are created should improve potential insights into plant metabolism from isotopic measurements. Additional reviews discussing compound-specific mechanisms and isotopic effects have been published on photosynthesis (Brugnoli & Farquhar, 2000), photorespiration (Ghashghaie et al., 2003), biogeochemistry (Hayes, 1993), and secondary metabolism (Schmidt & Gleixner, 1998; Hayes, 2002).

Most chemical and biochemical processes favor the initial incorporation of the lighter isotope in the product, leaving the substrate enriched in the heavy isotope. The partitioning of isotopes in reactions is termed ‘isotopic fractionation’; the magnitude of isotopic fractionation differs depending on the elements involved and the specific reaction mechanism. Isotopically fractionating processes therefore result in variation in the isotopic ratios between substrate and product. These ratios depend on the isotopic ratio of the substrate, the proportion of substrate transformed to product, and whether the system is open or closed. To illustrate how different isotopic patterns can be created, we show in Fig. 1a a simple two-reaction scheme in an open system for the reactions A → B and B → C + D, with the first reaction having an isotopic fractionation of Δ1 and the second reaction an isotopic fractionation of Δ2 (Fig. 1). As Fig. 1b shows, the initial product in a reaction will express the full fractionation factor, then as the reaction proceeds the isotopic composition of the product will gradually approach that of the initial reactant. Figure 1b also shows that the relative isotopic pattern among compounds A–D depends on the relative flux to the products of the two reactions, even though the actual reactions (and their associated isotopic fractionations) are identical. In this example, A could represent carbohydrates; B a metabolically important intermediate such as pyruvate; C respired CO2, and D acetyl-CoA, the building block for lipid synthesis. Thus, although an overall isotopic mass balance must be preserved, the removal of variable amounts of 13C-depleted CO2 in different reaction systems will result in shifts in the isotopic composition of the bulk material consisting of the remaining components. If C3 and C4 plants differ in their flux patterns through these two reactions, then isotopic shifts among the different components will also differ.

Figure 1.

Mode of action of isotope fractionation in reaction networks. (a) Schematic showing how isotopic fractionation affects δ13C of substrates and products of two reactions: (1) A → B; (2) B → C + D, with fractionation for the first reaction set at Δ1 and fractionation for the second reaction set at Δ2. (b) Pattern of carbon isotopes for A, B, and C + D in the above two reactions. The proportion of A reacting to form B is given as f1 and the that of B reacting to form C + D as f2. Given different values of f1 and f2 (inline image), the isotopic patterns among A, B, and C + D will shift, as given by A′, B′, and C′ + D′.

An important distinction in interpreting isotopic patterns is whether kinetic or equilibrium isotopic effects dominate in a reaction. In kinetic isotopic effects, the light isotope usually reacts more quickly than the heavy isotope, so that the resulting product is depleted in the light isotope relative to the substrate. In contrast, in equilibrium reactions the back-reaction from product to substrate also occurs, and therefore the kinetic isotopic effects for both forward and back reactions must be considered. In general, in equilibrium reactions the heavy isotope will concentrate in the product with the stronger bonds (Bigeleisen, 1965).

II. Methods and terminology

Isotopic ratios are usually reported on the Vienna–PDB scale and calculated as δ13C (‰) = (Rsample/Rstandard − 1) × 1000‰, with R = 13C/12C and with Rstandard set to 0.0112372 (molar ratio) (Werner & Brand, 2001). Isotopic fractionation during reactions is defined as α = Rsubstrate/Rproduct. For convenience, isotopic fractionations are more commonly reported as Δ values, with Δ reported in ‰. α is related to Δ by:

Δ = α − 1(Eqn 1)

Isotopic fractionation can be related to isotopic ratios through the following equation:

Δ = (δ13Csubstrate − δ13Cproduct)/(1 + δ13Cproduct) ≈ δ13Csubstrate − δ13Cproduct(Eqn 2)

Another measure often used is simply the isotopic difference between two pools, defined as:

Δ = δ13Cpool 1 − δ13Cpool 2(Eqn 3)

For the relatively small isotopic fractionations observed in the studies cited (most < 2‰) and the relatively small depletion in 13C of C3 relative to C4 source material (≈ 14‰, O’Leary, 1989), Δ = δ13Cpool 1 − δ13Cpool 2 is very similar to Δ = α − 1. We also define the difference between isotopic fractionation in C3 and C4 systems formally as:

image(Eqn 4)

For example, for the fermentation of glucose to ethanol, C3-derived glucose has a δ13C of −27‰ and the resulting ethanol has a δ13C of −29‰, therefore ΔC3 = 2‰. For the corresponding system with C4 glucose, the δ13C of glucose is −13‰ and the resulting ethanol has a δ13C of −14‰, therefore ΔC4 = 1‰ and ΔC3–C4 also = 1‰.

III. Results

1. Intramolecular 13C patterns in glucose

Intramolecular patterns of 13C can potentially provide information about mechanisms of formation of compounds (Abelson & Hoering, 1961; Hayes, 1993). It may accordingly be instructive to examine 13C patterns in glucose, the first persistent product of C3 and C4 photosynthesis. Although glucose formation during photosynthesis has been extensively studied, only a single study has compared intramolecular 13C distributions in C3 vs C4 glucose (Rossmann et al., 1991). Individual atoms within glucose varied up to 10‰, with the C-4 atom most enriched in 13C and the C-6 atom most depleted in 13C (Table 1). Relative to the mean for each glucose type, the C-3 atom was enriched by 2.6‰ in C3 glucose relative to C4 glucose, and the C-1 atom was enriched in C4 glucose relative to C3 glucose by 2.2‰ (Table 1).

Table 1.  Intramolecular isotopic pattern of 13C in C-1 to C-6 of glucose relative to average isotopic composition reported by Rossmann et al. (1991) from fermentation analysis, and partial data from Ivlev et al. (1987) (I) and Gleixner et al. (1993) (G)
PositionΔ (C3)Δ (C4)Δ (C3–C4)Δ (C4) (I)Δ (C3) (G)
  1. Values in ‰, with Δ calculated as δ13Cposition − δ13Caverage. Calculated isotopic differences (Δ) for C-1 to C-6 positions of glucose relative to the bulk value are also given. aCalculated. bC-3 and C-4 measured together.

C-1−1.3 0.9−2.2−2.35−2.6a
C-2−0.9−0.1−0.8 1.7a−2.6a
C-3 1.9−0.7 2.6 1.7a 5.2b
C-4 6.3 5.2 1.1 1.7a 5.2b
C-5−1.1−0.1−1.0 1.7a−2.6a

2. Biochemical transformations

In simple, two-component biochemical reactions in which both substrate and product are isotopically characterized, the δ13C of products should reflect the isotopic ratios of atoms at specific intramolecular positions of the substrate that are transferred to the product, plus any isotopic fractionation associated with biochemical reactions. If differences in intramolecular 13C distributions are common in C3 and C4 plants, then evidence of such differences should appear whenever subsequent biochemical transformations result in preferential retention or loss of carbon atoms from specific positions within molecules.

Fermentation  Fermentation experiments are valuable probes of intramolecular isotopic patterns because they are position-specific, with ethanol produced during fermentation by Saccharomyces cerevisiae derived from the C-1, C-2, C-5 and C-6 positions within hexoses. With a high yield, the reaction is not isotopically fractionating (Scrimgeour et al., 1988; Weber et al., 1997; Zhang et al., 1998). Therefore the evolved CO2 should reflect the isotopic compositions of the C-3 and C-4 positions within hexoses, and the ethanol should reflect the isotopic compositions of the C-1, C-2, C-5 and C-6 positions (Fig. 2). For example, Weber et al. (1997) fermented C3 glucose in which the intramolecular isotopic pattern had previously been determined by Rossmann et al. (1991). The 13C depletion in ethanol and 13C enrichment in CO2 were exactly that predicted from knowledge of the intramolecular isotopic composition of glucose. As a result of differing 13C distributions of C3 and C4 glucose, the isotopic shift between glucose and ethanol during fermentation should differ for C3 and C4 glucose.

Figure 2.

Schematic of glucose fermentation products. Numbering shows carbon atom source from parent glucose molecule for ethanol and CO2.

C3 and C4 systems clearly differed in patterns of isotopic fractionation in several fermentation experiments. The 13C depletion relative to the parent carbohydrate for C3-derived ethanol ranged from 1 to 2‰, whereas the depletion for C4-derived ethanol ranged from 0.3 to 1.2‰ (Table 2). For the fermentation of sucrose to ethanol, calculated values of ΔC3–C4 were 0.8 and 1.0‰ (Table 2). The CO2 evolved during yeast fermentation derives from the C-3 and C-4 atoms within glucose. CO2 evolved during yeast fermentation of C3 and C4 glucose was 7.4‰ (C3) and 5.1‰ (C4) enriched relative to the source glucose (Scrimgeour et al., 1988). The value for ΔC3–C4 of −2.3‰ was between calculated ΔC3–C4 values of −1.8‰ for the C-3 and C-4 atoms from Rossmann et al. (1991) and −2.9‰ using C3 data from Gleixner et al. (1993) and C4 data from Rossmann et al. (1991).

Table 2.  Isotopic patterns during fermentation of saccharides to ethanol and CO2
MeasurementC3 (‰)C4 (‰)ReferenceComments
Sucrose–ethanol 1.8 0.8Weber et al. (1997) 
Sucrose–ethanol 2 1.2Zhang et al. (1998) 
Starch–ethanol 1.7 0.3, 0.3Rauschenbach et al. (1979)Potato, corn
Glucose–ethanol 2.0 Weber et al. (1997)Sugar beet
Glucose–ethanol 1.5 Deléens et al. (1990) 
Glucose–ethanol 1.6 Zhang et al. (1991)Potato
Sugars–ethanol 1.6 ± 0.2 Rossmann et al. (1996)From grape must
Glucose–CO2−7.4 ± 0.1−5.1 ± 0.03Scrimgeour et al. (1988)n = 3
Glucose–CO2−4.6 ± 0.1 Weber et al. (1997) 

Respiration  Isotopic fractionation during autotrophic and heterotrophic respiration is of interest at both global and local scales because the δ13C of soil-respired and atmospheric CO2 is used to partition sources and sinks of CO2 (Amundson et al., 1998). In global models of carbon budgets, isotopic fractionation during C3 and C4 respiration is assumed to be identical. In a study of respiration by isolated plant mesophyll protoplasts, Lin & Ehleringer (1997) concluded that isotopic fractionation during dark respiration of C3 and C4 plant protoplasts when fed C3 or C4 sucrose and glucose was identical and negligible. Given that metabolic pathways in isolated plant protoplasts are somewhat restricted, a simple explanation for their results is that, under their conditions, possibilities for branch-point reactions and efflux of carbon out of protoplasts were minimized (Ghashghaie et al., 2003). As a consequence, any carbohydrate taken up may have been completely metabolized to CO2, thereby eliminating any opportunity for isotopic fractionation to be expressed.

The evidence for no isotopic fractionation during respiration disagrees with leaf-level studies showing differences in isotopic fractionation during respiration for C3 and C4 plants (Duranceau et al., 1999; Duranceau et al., 2001; Ghashghaie et al., 2001). In these studies, CO2 released during dark respiration in C3 plants was 0–6‰ enriched in 13C relative to the presumed respiratory substrates of glucose and sucrose, whereas CO2 enrichment in a C4 plant was ≈ 1‰ (J. Ghashghaie, unpublished). These patterns were similar to those reported by Rooney (1988) comparing respired CO2 and bulk leaf matter in soybeans (5‰ enrichment of CO2) and corn (1‰ enrichment of CO2).

3. Isotopic patterns among components of C3 and C4 plants

Differences in bulk δ13C of different plant parts (e.g. leaves, roots) are common. Plant roots are almost invariably enriched relative to leaves in C3 plants by 1–3‰ (Table 3), whereas roots in C4 plants appear to be similar or slightly lower in δ13C relative to leaves. Grains from C3 plants were also enriched by 1–4‰ relative to leaves, whereas C4 grains were ≈ 1.5‰ enriched relative to leaves in maize (Table 3).

Table 3.  Isotopic differences between leaves and other plant components
SpeciesValue (‰)Reference
Grain − leaves (C3)
 Oats3.3, 3.6Winkler et al. (1978)
 Wheat1.2, 2.0Yoneyama et al. (1997)
 Wheat2.9, 4.6Winkler et al. (1978)
 Legumes (7 species)0.9 ± 0.2Yoneyama & Ohtani (1983)
Grain − leaves (C4)
 Maize1.5Lowdon (1969)
 Maize1.4Gleixner et al. (1993)
Roots − leaves, woody vegetation (C3)
 Citrus2.4Syvertsen et al. (1997)
 Eucalyptus0.9Handley et al. (1993)
 Fagus1.5Fotelli et al. (2003)
 Pinus1.5E. Hobbie & J. Colpaert (unpublished)
 Populus2.5Ineson et al. (1996)
 Pseudotsuga1.8Hobbie et al. (2002)
Roots − leaves, nonwoody vegetation (C3)
 Barley0.4 ± 0.2Hubick & Farquhar (1989)
 Barley − stressed0.5 ± 0.3Hubick & Farquhar (1989)
 Beta (sugar beet)1.2Gleixner et al. (1993)
 Desmodium1.5Schweizer et al. (1999)
 Legumes (seven species)1.0 ± 0.1Yoneyama & Ohtani (1983)
 Peanut0.5Hubick et al. (1986)
 Plantago1Staddon et al. (1999)
 Ricinus0.6Handley et al. (1993)
 Tomato1.1Bradford et al. (1983)
 Wheat1.2 to 1.4Cheng & Johnson (1998)
 Wheat1.8Lichtfouse et al. (1995)
Roots − leaves (C4)
 Four species−0.2 to 0.2Trouve et al. (1994)
 Brachiaria−0.5Schweizer et al. (1999)
 Pennisetum−1.8DesJardins et al. (1994)
 Saccharum 0.4Spain & Le Feuvre (1997)
 Zea−1.4Qian & Doran (1996)

Differences in δ13C among plant compounds are also common (Table 4). In δ13C analyses in bulk leaf matter, lipids, and alkanes that included C3 and C4 plants, Collister et al. (1994) reported that the general order for δ13C was alkanes < lipids < bulk. Relative to bulk values, alkanes and lipids were 4–6‰ depleted in C3 plants and 8–10‰ depleted in C4 plants. Chikaraishi & Naraoka (2001) recorded similar patterns, and also reported that the concentration of lipids in C4 plants was about half that in C3 plants. However, in a study of grassland vegetation Conte et al. (2003) concluded that C3 and C4 plants did not differ in 13C depletion of lipids relative to bulk. The well known isotopic enrichment of plant cellulose relative to lignin is apparently somewhat greater in C4 than in C3 leaves, with ΔC3–C4 equal to −2.1‰ (Benner et al., 1987) or −0.5‰ (Schweizer et al., 1999), with comparable values in roots (ΔC3–C4 = −2.1‰, Schweizer et al., 1999).

Table 4.  Isotopic patterns among plant compounds arising from different photosynthetic pathways
MeasurementC3 (‰)C4 (‰)ReferenceComments
  1. Values are from leaves unless otherwise noted.

Cellulose − lignin3.5 Wilson & Grinsted (1977)Wood
Cellulose − lignin4.1 ± 0.6 6.2 ± 0.9Benner et al. (1987)Various tissues
Cellulose − lignin4.6 5.1Schweizer et al. (1999)Leaves
Cellulose − lignin2.5 4.6Schweizer et al. (1999)Roots
Bulk − alkanes5.9 9.9Collister et al. (1994) 
Bulk − alkanes3.3 8.5Chikaraishi & Naraoka (2001) 
Bulk − C27 n-alkane2.7 ± 0.1 Lockheart et al. (1997)Fagus sylvatica
Bulk − n−alkanols5.4 ± 1.0 9.5Conte et al. (2003)Native grasses
Bulk − n−alkanoic acids5.0 ± 0.510.1Conte et al. (2003)Native grasses
Bulk − n−alkanes6.4 ± 0.2 9.1Conte et al. (2003)Native grasses
Bulk − n−lipids8.8 ± 0.3 (n = 4) Conte et al. (2003)Crops
Bulk − lipids1.9–8.5 (n = 4) Park & Epstein (1961) 
Bulk − lipids9 Smith & Jacobson (1976)Potato tubers
Bulk − lipids4.3 Winkler et al. (1978)Oat
Bulk − lipids1.9 6.6 ± 1.1Ballentine et al. (1998) 
Bulk − fatty acids6.5 ± 0.610.3 ± 0.5Ballentine et al. (1998)C14–C22 fatty acids

IV. Discussion

1. Mechanisms

The examples given from the literature suggest that isotopic differences between C3- and C4-derived substrates are common at all scales. Several differences in metabolic pathways between C3 and C4 plants could contribute to these differences, including photorespiration, sugar metabolism and fluxes during the reductive pentose phosphate pathway.

Aldolase, triose phosphate isomerase and the reductive pentose phosphate cycle  To understand 13C patterns in C3- and C4-derived glucose, we must understand how carbon cycles during enzymatic reactions of photosynthesis and glucose synthesis, and how those enzymatic reactions fractionate against 13C. One hypothesis was advanced by Gleixner & Schmidt (1997), who proposed that equilibrium isotopic fractionation during the formation of fructose-1,6-bisphosphate via aldolase from dihydroxyacetone phosphate (DHAP) and glyceraldehyde-3-phosphate (GAP) partly controlled the 13C enrichment at the C-3 and C-4 positions of glucose. However, the measured equilibrium isotopic effects would create enrichments of only 1‰ for C-4 relative to C-3, so additional mechanisms must be invoked to account for the large measured enrichments between these two positions (Table 1; 4‰ for C3 glucose, 6‰ for C4 glucose).

Cycling through hexose monophosphates and triose phosphates may influence 13C content by allowing the exchange of specific positions within molecules. For example, such cycling in oak stems resulted in ≈ 25% of C-1 labeled glucose subsequently appearing at the C-6 position of fructose, indicating that 50% of glucose had undergone exchange (Hill et al., 1995). Similar cycling levels (≈ 40% exchange) were reported from cell suspensions of carrot (Krook et al., 2000). When coupled to glycolysis, triose phosphate isomerase will isomerize the two halves of the glucose molecule. Such isomerizations should decrease isotopic differences between the two exchanged positions, unless the isomerization reaction itself is isotopically fractionating. Despite the central importance of the triose phosphate isomerase reaction in glycolysis, isotopic fractionation during this reaction is currently unknown.

The intensity of short-term, position-specific labeling in glucose of plants exposed to 14CO2 (Gibbs & Kandler, 1957) is highly correlated with the natural abundance 13C pattern in C3 glucose (Fig. 3). The reasons for this correspondence are unclear. One possibility is that small isotopic fractionations during regeneration of ribulose-1,5-bisphosphate (RuBP) (the reductive pentose phosphate or Calvin cycle) and during regeneration of phosphoglycerate from glycolate (the photorespiratory cycle) result in 13C depletions roughly proportional to the number of times carbon has cycled through these two pathways. Under this scenario, the C-4 atom is most enriched in 13C because it has undergone, on average, the fewest potentially fractionating steps since the initial incorporation as CO2. Additional short-term labeling studies suggest that glucose is highly labeled in 14C at C-4 because pre-existing pools of sugars or triose phosphates preferentially dilute DHAP that forms the C-1 to C-3 positions of glucose (Trebst & Fiedler, 1962), whereas newly incorporated CO2 preferentially resides at C-1 of GAP that will form the C-4 to C-6 positions of glucose. This pattern presumably arises because the free energy of the GAP → DHAP equilibrium strongly favors DHAP (20 : 1, Gibbs free energy −7.5 kJ mol−1, Zubay, 1984), and the GAP + DHAP → fructose-1,6-bisphosphate equilibrium even more strongly favors fructose-1,6-bisphosphate formation (Gibbs free energy −23.9 kJ mol−1). As a consequence, newly created GAP is preferentially incorporated into fructose-1,6-bisphosphate (vs isomerization to DHAP), and little unlabeled DHAP isomerizes to GAP. Therefore GAP becomes labeled with 14C more quickly than DHAP (Fig. 4). The smaller intramolecular 13C shifts in C4 glucose may reflect in part a lack of rearrangements caused by photorespiration. Unfortunately, short-term labeling data to test this hypothesis are unavailable for C4 glucose.

Figure 3.

Patterns of short-term labeling with 14C in glucose of different photosynthetic organisms correlate with natural abundance 13C labeling. Time after 14C introduction when photosynthesis was quenched is shown. Correlations (r2) between 14C and natural abundance 13C measurements on C3 glucose (Δ from Table 1) are shown in parentheses below species designations.

Figure 4.

Hypothesized pattern of 14C-labeled CO2 incorporation into RuBP and subsequent transformations leading to hexose formation. The Gibbs free energy of aldolase and triose phosphate isomerase reactions, coupled with large pre-existing pools of DHAP, result in preferential incorporation of newly incorporated 14C at C-4 of glucose. Steps after fructose-1,6-bisphosphate formation are not shown.

Photorespiration  One of the most plausible mechanisms that could produce differences in 13C patterns within C3- and C4-derived sugars is photorespiration, an important reaction in C3 metabolism but not in C4 metabolism (Edwards & Walker, 1983). During photorespiration, RuBP is split into 3-phosphoglycerate and 2-phosphoglycolate. Two glycolate molecules are subsequently reincorporated into glycerate with the loss of one carbon dioxide molecule. After phosphorylation, the 3-phosphoglycerate produced becomes available for sugar synthesis. Glycerate derived from glycolate therefore consists of two carbons from the C-1 position of RuBP and one carbon from the C-2 position of RuBP (Fig. 5). Because of differing rates of incorporation of the individual carbons of RuBP into glucose between C3 and C4 photosynthetic pathways, and the numerous opportunities for isotopic fractionation during RuBP synthesis (Rossmann et al., 1991) and during photorespiration itself (Ivlev, 1993), different δ13C values could easily be generated. The oxygenase activity of ribulose bisphosphate carboxylase-oxygenase (Rubisco) may be as much as 40% that of Rubisco carboxylase activity in C3 plants (Douce & Heldt, 2000), with one triose created for every two oxygenation reactions, and one triose exported for every three carboxylation reactions. Therefore the maximum proportion of initially fixed carbon that passes through the photorespiratory and glyoxylate cycles is calculated to be as much as (40% × 1/2 + 100% × 1/3)/(40% + 100%), or 37.5%. As photorespiration creates different combinations of the constituent RuBP atoms in the trioses generated by photosynthesis, any isotopic differences among the RuBP atoms lead to different intramolecular isotopic patterns in C3 and C4 sugars derived from these trioses.

Figure 5.

Simplified schematic of the photorespiratory cycle showing how atoms derived from RuBP are transferred to phosphoglycerate after photorespiration.

The degree to which 13C carbon is fractionated during photorespiration has not been extensively investigated, primarily because of difficulties in separating total respiration in the light into photorespiration and dark respiration. A value of 8‰ for photorespiration is used in most models of photosynthesis that include isotopic effects (Lloyd & Farquhar, 1994; although the cited source, Rooney, 1988, actually reports values of 6.2 and 7.4‰). Gillon et al. (1998) obtained values of 3.3 ± 1.6‰ (wheat, Triticum aestivum) and close to 0‰ (bean, Phaseolus vulgaris). However, later unpublished results by Gillon et al. on the same plants (cited by Brugnoli & Farquhar, 2000) indicated isotopic fractionation during photorespiration closer to the original Rooney (1988) values. A possible confounding factor in these studies is the potentially high degree of refixation of respired and photorespired CO2, estimated at over 80% (Loreto et al., 1999). In addition, the ratio of photosynthesis to photorespiration may change as a result of various stresses on plants, particularly water stress (Douce & Heldt, 2000). These stresses may result in altered intramolecular isotopic patterns because of additional isotopic fractionation during photorespiration and concurrent changes in the flux patterns of carbon from trioses to glucose.

Isotopic fractionation during the CO2-releasing step of photorespiration has been most extensively studied by Ivlev and colleagues (Ivlev et al., 1996; 1999; Igamberdiev et al., 2001). Their studies of the reaction mechanism of isotopic fractionation by glycine decarboxylase (the enzyme responsible for CO2 release during photorespiration) indicate that isotopic fractionation during photorespiration varies with reaction pH, supply of various cofactors, and plant species. Across seven different species, isotopic fractionation averaged 7.6 ± 1.7‰ after 2 h, decreasing to 5.3 ± 2.4‰ after 16 h. Indirect support for fractionation during photorespiration is also provided by δ13C patterns in amino acids derived from C3 and C4 plants (Fogel & Tuross, 1999). Of 10 amino acids sampled, the average difference between C3 and C4 acids was 14‰, with the smallest difference in glycine, only 9‰. Fractionation during photorespiration against 13C should enrich the 13C content of C3 glycine, but not that of C4 glycine, leading to a smaller 13C enrichment of C4- vs C3-derived glycine compared with other amino acids.

Branch points  The relatively small isotopic differences between C3 and C4 glucose in 13C intramolecular patterns can account for δ13C patterns in fermentation products. However, 13C patterns in glucose appear inadequate for explaining the large differences in 13C content reported between bulk and lipids, or between bulk components, because (1) the largest position-specific difference of glucose is only 2.6‰; and (2) overall patterns of carbon incorporation are relatively well balanced among the different carbon atoms of glucose, so that overall fractionation based solely on differential incorporation of glucose atoms is close to 0‰. For example, in a study of the retention by cultured fungi of position-specific, 13C-labeled glucose, the average relative incorporation of the C-1 through C-6 positions was 14, 20, 18, 9, 18 and 21%, respectively (E. Hobbie and co-workers, unpublished). Based on the 13C patterns for glucose in Table 1, this translates into a fractionation against bulk glucose values of only 0.3‰ for C3 glucose and 0.2‰ for C4 glucose, for a difference of only 0.1‰ between the two glucose types. We conclude that isotopic fractionation during postphotosynthetic reactions must cause most isotopic differences between components at the bulk- and compound-specific levels.

Isotopic effects in reactions arise through a combination of isotopic fractionation and metabolic branch points (Schmidt & Gleixner, 1998). Accordingly, how C3 and C4 plants allocate carbon along various metabolic pathways will influence patterns of isotopic fractionation during reactions. Some of the most important enzymatic reactions influencing isotopic composition are listed in Table 5 along with measured fractionation factors. Given the comparatively recent evolution of the C4 photosynthetic pathway, isotopic fractionations during most enzymatic reactions are probably similar for C3 and C4 plants. If so, isotopic differences for these two photosynthetic pathways in intramolecular patterns or between different plant components must arise primarily through differences in proportions of products proceeding through alternative reaction pathways coupled to isotopic fractionation during reactions.

Table 5.  Selected enzyme-catalyzed reactions with measured kinetic carbon isotope effects that could influence the inter- and intramolecular carbon isotope distribution in primary and secondary plant metabolites
Reaction with identified kinetic isotope effectEnzyme (trivial name)Consequence on intramolecular 13C pattern or bulk δ13C valueCarbon atomEffect (‰)Reference
RuBP + CO2→ 2 PGARubisco13C-depletion of C3 plants relative to ambient CO229(1, 2)
PEP + HCO3→ HO2C-CH2-CO-CO2H + PPEP carboxylase13C-depletion in C4 plants; relative 13C-enrichment in position C-4 of oxaloacetate; 13C-enrichment in position C-4 of oxaloacetate and descendants in C3 plants2(1, 2) (3, 4)
HO2C-CH2-CO-CO2H → H3C-CO-CO2H + CO2Pyruvate carboxylase13C-enrichment in position C-4 of organic acids 13C-depletion in C-3 of organic acids due to 13C-depletion in C-1 and C-6 in glucoseC-424, 7.5a(5)
Fru-1,6-BP → GAP + DHAPFructose-1,6-bisphosphate aldolase13C-enrichment in position C-3 and C-4 of carbohydrates, depending on position of equilibrium (catabolic or anabolic metabolism)C-316 ± 7, 4a(6)
C-4−3 ± 9, 5a 
Glucose-6-phosphate → 6-phosphogluconateGlucose-6-phosphate dehydrogenaseDepletion of C-1 of 6-phosphogluconateC-116.5, −8a(7)
6-phosphogluconate → CO2+ ribulose-5-phosphate6-phosphogluconate dehydrogenase13C-depletion of CO2 respired in pentose phosphate cycle; enrichment of C-1 of 6-phosphogluconateC-110, 21(8, 9)
Decarboxylation of malateMalic enzyme13C-enrichment of acetyl group in acetyl CoA (C-2 of malate); 13C-depletion of methyl group in acetyl-CoA (C-3 of malate); 13C-depletion of fatty acidsC-11(1)
H3C-CO-CO2H + HSCoA → H3C-CO-SCoA + CO2+ 2 [H]Pyruvate dehydrogenase13C-depletion in CO group of acetyl-CoA, 13C-depletion of fatty acids and isoprenoids, alternating 13C-pattern in some lipidsC-19, 24b(10)
C-221, 25b 
C-33, 3b 
H3C-CO-CO2H → H3C-COH + CO2Pyruvate decarboxylase13C-depletion in acetaldehyde and descendants, 13C-depletion in carbonyl group of acetaldehydeC-17, 8, 10c(11–13)d
C-215, 13c 
2 glycine → CO2 + serineGlycine decarboxylasePhotorespiration; relative 13C-depletion in CO2 with a corresponding enrichment in serineC-18 ± 7e(14)

The relative allocation to 13C-enriched vs 13C-depleted compound classes should presumably also influence 13C patterns of specific compound classes relative to bulk tissue. Because an isotopic mass balance must be preserved, species with relatively small allocations to 13C-depleted compound classes, such as lignin or lipids, should show greater 13C depletion in these classes relative to bulk δ13C than species with larger allocations to 13C-depleted compound classes. This simple yet powerful idea was invoked by Park & Epstein (1961) to explain the increased 13C depletion in lipids relative to bulk with decreased lipid content of marine and terrestrial autotrophs, and appears generally applicable to explaining δ13C patterns in lipids of higher plants (Fig. 6). The offset in Fig. 6 of trees relative to other plants presumably arises because of high concentrations of 13C-depleted lignin in these samples; the correlation without trees has an r2 of 0.83, and the correlation of data from the four trees is 0.95. These results indicate that the greater 13C depletion of lipids relative to bulk in C4 than in C3 plants, reported in several studies, primarily derives from the lower lipid and lignin content in C4 grasses than in C3 woody vegetation. The driving mechanism illustrated in Fig. 7a could therefore also explain the lesser 13C depletion in lignin relative to cellulose in C3 woody vegetation than in C4 grasses (Table 4; Fig. 8), as tissues with lower lignin concentrations generally show greater 13C depletion in lignin relative to cellulose or bulk tissue. A survey of published reports indicates that high concentrations of lignin and lipids are common in woody tissue or foliage of woody plants, with little differentiation between C3 and C4 grasses in lignin and lipid concentrations (Table 6; Fig. 7b). Thus the high depletion in 13C of lipids relative to bulk carbon is a property of many grasses and herbaceous plants.

Figure 6.

Correlations between lipid concentrations in foliage and the isotopic depletion of lipids relative to bulk carbon differ for tree and nontree vegetation. Open circles, trees; filled circles, C4 grasses; filled squares, C3 herbaceous vegetation; upward-pointing triangle, fern; downward-pointing triangle, shrub. Data from Park & Epstein (1961) and Chikaraishi & Naraoka (2001). The correlation (r2) for nontrees is 0.83 (n = 9) and for trees is 0.95 (n = 4), with the overall correlation 0.47.

Figure 7.

Interaction of synthesis reactions and transport processes influences isotopic composition of mobile and immobile compounds in different plant tissues. (a) Movement and isotopic fractionation of carbon between plant leaves and roots in different compounds. Isotopic fractionation during synthesis of lignin and lipids results in 13C-depleted products and 13C enrichment in residual substrates such as sugars, which are then transported to roots. (b) Isotopic depletion of lignin and lipids depends on the fraction (f) of available substrate transformed to lignin and lipids and the isotopic fractionation (Δ) of the reaction. Thus the smaller value of f for grasses than for trees results in greater 13C depletions of lignin and lipids relative to bulk δ13C in grasses.

Figure 8.

Lignin concentrations in different tissues are correlated with the isotopic depletion of lignin relative to cellulose (r2 = 0.37, n = 19, P = 0.004). Data from Benner et al. (1987) and Fernandez et al. (2003). Open circles, trees; filled circles, C4 grasses; open squares, C3 herbaceous vegetation.

Table 6.  Concentrations of 13C-depleted compound classes in C3 and C4 plants
Plant type and tissueLipidsWaxesLigninTannins or PhenolicsReference
  1. Concentrations in ‰. Error bars represent standard errors. Plants are C3 unless indicated.

C3 wood (n = 4)  251 ± 36 Benner et al. (1987)
C3 rhizome (Juncus)  51 Benner et al. (1987)
C3 leaf (n = 2) (Juncus, Carex)  44–60 Benner et al. (1987)
C4 leaf (n = 3)  61 ± 11 Benner et al. (1987)
C4 rhizome (Spartina)  93 Benner et al. (1987)
Coniferous forest18 305 Poorter (1989)
Deciduous forest19 208 Poorter (1989)
Lepechinia calycina (shrub)134 210 Poorter (1989)
Senecio jacobea (roots)25  80 Poorter (1989)
Cucumis sativis (leaves)41 26 Poorter (1989)
C. sativus (stems)26 42 Poorter (1989)
C. sativus (roots)27 40 Poorter (1989)
Zea mays (C4)25 80 Poorter (1989)
Sorghum (C4), above-ground30 27 Poorter (1989)
C3 crop (n = 9)53 ± 2 28 ± 612 ± 2Poorter et al. (1997)
C3 wild herbaceous (n = 11)50 ± 4 40 ± 618 ± 3Poorter et al. (1997)
C3 woody vine (Vitus)59 5734Poorter et al. (1997)
C3 trees (n = 6)82 ± 11 76 ± 568 ± 14Poorter et al. (1997)
C3 grass (Holcus)  67 Ross et al. (2002)
C4 grass (Pennisetum)  89 Ross et al. (2002)
Trees (roots)2414215183Martinez et al. (2002)
Shrubs (roots)2713912945Martinez et al. (2002)
Grasslands (roots)14 40165 8Martinez et al. (2002)

We suggest that greater allocation of carbon to lignin and lipid pools in woody plants than in nonwoody plants could largely account for the greater 13C enrichment of roots vs leaves in woody C3 plants than that in nonwoody C3 or C4 plants (Table 3). As a consequence of this allocation, the remaining carbohydrates that transport carbon to roots are more enriched in 13C relative to bulk pools in woody than in nonwoody plants, leading to greater relative 13C enrichment in roots relative to foliage in woody than in nonwoody plants. Thus, many isotopic differences between different compound classes and between plant pools along transport pathways may arise from a single, unifying mechanism.

2. Implications and future research

We have posited here that many previously reported 13C differences between compound classes or different plant pools of C3 or C4 plants arise primarily from differences between woody and herbaceous vegetation in their relative allocation to 13C-depleted biosynthetic pathways leading to 13C-depleted lignin or lipids, not because of intramolecular 13C distributions in sugars created during C3 and C4 photosynthesis. The general applicability of this conclusion could be tested by measuring δ13C patterns in foliage, roots and wood of woody C4 plants, for which only foliar δ13C measurements currently exist (Pearcy & Troughton, 1975). Future work should also stress comparative studies of variations in concentration and δ13C signatures of compound classes across a range of C3 and C4 species and tissues, as such studies should reveal how concentrations and δ13C signatures covary.

Because of the potential insights hidden within position-specific 13C patterns of biochemical mechanisms of formation, improvements in measuring such patterns are needed. Position-specific measurements have usually required careful enzymatic and chemical degradation of compounds in multiple reactions to examine isotopic patterns in the resulting products. The labor-intensive nature of such work may explain why the seminal work of Rossmann et al. (1991) on intramolecular patterns in glucose has yet to be repeated. Nuclear magnetic resonance methods are potentially powerful sources of position-specific information, but their use in detecting isotopic ratios at natural abundance levels has been mainly restricted to compounds of two or three carbon atoms (Zhang et al., 1998). An alternative approach, requiring pyrolysis of compounds and isotopic characterization of the resulting molecular fragments, holds considerable promise for deducing 13C distributions within compounds (Brenna et al., 1998). Such studies could establish fragmentation patterns by pyrolyzing compounds that are 13C-labeled at specific intramolecular positions, followed by comparison with pyrolysis of natural abundance compounds to determine intramolecular 13C distributions. Current studies suggest that differences between C3 and C4 glucose in intramolecular isotopic patterns could be detected in this way (G. Gleixner, unpublished).

Computer modeling is a potentially valuable tool to explain isotopic patterns at different scales. The recent development of models for understanding the metabolic fate of 13C-labeled compounds in autotrophic and heterotrophic tissues of plants suggests that these models could be adapted to understand natural 13C distributions as well. Their application was recently reviewed by Roscher et al. (2000). Similar modeling efforts in bacteria, aimed at improved industrial production of biochemicals, have incorporated all the primary reactions of cell metabolism and have successfully predicted the metabolic pathways of position-specific labeled compounds (Schmidt et al., 1999). In contrast to modeling of 13C tracers, natural abundance modeling is limited to Ivlev's (1985) work on position-specific 13C distributions during plant metabolism. Analyses of position-specific potential isotopic fractionations during photorespiration, photosynthesis and sugar synthesis have yet to be attempted, and computer modeling of secondary metabolic pathways such as lignin and lipid synthesis is also unexplored. Such modeling would undoubtedly provide many new insights into the creation and interpretation of intramolecular and compound-specific isotopic patterns, particularly once 13C fractionations were determined for key enzymes, such as triose phosphate isomerase, for which we currently lack information. Results from these modeling efforts could then be readily applied to understanding bulk isotopic signatures of plant components.

At the larger scale of modeling bulk isotopic signatures of plant components, our recent work indicates that 13C increases between foliage and roots in trees could plausibly be explained by either discrimination against 13C during respiration, or discrimination against 12C during creation of transfer compounds such as sucrose (E.H. and J. Colpaert, unpublished). Future modeling should explore why shoot-to-root 13C increases are smaller in C3 herbaceous plants, and appear to disappear altogether in C4 plants. Such modeling should help resolve the important issue of 13C discrimination during respiration, and how the results of Lin & Ehleringer (1997) showing no discrimination in plant protoplasts can be reconciled with leaf-based measurements showing variable levels of discrimination.

V. Conclusions

Differences in isotopic patterns between C3 and C4 plant components may derive from various causes. Differences in intramolecular isotopic patterns could affect isotopic patterns at both compound-specific and bulk levels. At the intramolecular level, photorespiration and alterations in RuBP regeneration caused by photorespiration appear the strongest candidates for the root causes of isotopic differences. Intramolecular isotopic patterns may reveal relative levels of photosynthesis and photorespiration in C3 and C4 plants, particularly if studied through computer modeling of natural and tracer isotope distributions during plant metabolism. Differences between C3 and C4 plants in allocation to mobile vs immobile compound classes probably influence isotopic differences between specific compounds or bulk plant pools more than intramolecular isotopic differences. Future studies using C3 and C4 substrates to study dietary preferences or soil carbon turnover should consider possible effects of nonhomogeneous 13C patterns between substrates.


Comments by John Hayes, Andreas Rossmann, Hanns-Ludwig Schmidt, Leo Sternberg and two anonymous reviewers greatly improved earlier drafts of this paper. The assistance of Galina Churkina in translating Russian articles and the support of the Max Planck Society and the US National Science Foundation (grant DEB-0235727) are gratefully acknowledged.