Why are Nitrogen Concentrations in Plant Tissues Lower under Elevated CO2? A Critical Examination of the Hypotheses

Authors


  • Supported by the Cullen Fund of Southwestern University to D. R. Taub.

*Author for correspondence.
Tel: +1 512 863 1583;
Fax: +1 512 863 1696;
E-mail: <taubd@southwestern.edu>.

Abstract

Plants grown under elevated atmospheric [CO2] typically have decreased tissue concentrations of N compared with plants grown under current ambient [CO2]. The physiological mechanisms responsible for this phenomenon have not been definitely established, although a considerable number of hypotheses have been advanced to account for it. In this review we discuss and critically evaluate these hypotheses. One contributing factor to the decreases in tissue N concentrations clearly is dilution of N by increased photosynthetic assimilation of C. In addition, studies on intact plants show strong evidence for a general decrease in the specific uptake rates (uptake per unit mass or length of root) of N by roots under elevated CO2. This decreased root uptake appears likely to be the result both of decreased N demand by shoots and of decreased ability of the soil-root system to supply N. The best-supported mechanism for decreased N supply is a decrease in transpiration-driven mass flow of N in soils due to decreased stomatal conductance at elevated CO2, although some evidence suggests that altered root system architecture may also play a role. There is also limited evidence suggesting that under elevated CO2, plants may exhibit increased rates of N loss through volatilization and/or root exudation, further contributing to lowering tissue N concentrations.

Growth of plants at atmospheric concentrations of carbon dioxide (CO2) greater than the current ambient can greatly affect plant tissue chemistry (Poorter et al. 1997; Loladze 2002). One of the most commonly seen effects is a decrease in the dry mass concentration of N (Nm). Cotrufo et al. (1998), synthesizing data from a broad range of studies, found mean decreases in Nm of 14% in aboveground tissues and 9% in roots. This compares closely with the findings of other data syntheses, which have found elevated [CO2] mediated decreases in Nm of 16.4% for leaves of woody plants (Curtis and Wang 1998), 11% and 14% respectively for leaves of gymnosperms and angiosperms in open-top chamber experiments (Norby et al. 1999), 12.9% for leaves in free-air carbon dioxide enrichment (FACE) experiments (Ainsworth and Long 2005), 14% for seeds (Jablonski et al. 2002) and 9–15% for the edible portions of several major food crops (although only 1.4% for soybean; Taub et al. 2008). Such decreases in Nm can have important implications not only for plant physiological processes, but for food chains, as the performance of insect herbivores often decreases with decreases in Nm (Bezemer and Jones 1998; Zvereva and Kozlov 2006), and changes in plant tissue quality under elevated CO2 can affect herbivore population dynamics (Whittaker 1999).

While broad literature surveys have consistently found mean decreases in Nm of approximately 10–15% for plants grown at elevated [CO2], there is considerable variability in the results of individual studies. Yin (2002) found that, across a wide range of studies, the elevated [CO2] effect on leaf Nm ranged from a 56% decrease to a 20% increase, while Norby et al. (1999) found a range from a decrease of 35% to an increase of 20% for the leaves of trees grown in open-top chambers. A number of factors have been identified that partially explain this variation among studies. Yin (2002) found that the effect of elevated [CO2] was greatest for woody deciduous species and that decreases in Nm under elevated CO2 were most pronounced at high light levels, high temperatures and large pot sizes. Yin (2002) and Taub et al. (2008) both found that the effect of [CO2] on Nm was reduced by N fertilization. Several studies have reported that the effect of elevated [CO2] on Nm is less for nitrogen-fixing species than for other types of plants (Cotrufo et al. 1998; Jablonski et al. 2002; Taub et al. 2008). Ainsworth and Long (2005) and Taub et al. (2008) found that the effect of elevated CO2 on Nm increased under ozone stress (although see Taub et al. 2008 for divergent results for soybean).

Although it has been well established that elevated [CO2] typically decreases Nm, the mechanisms by which this occurs are not certain, although a large number of hypotheses have been advanced to explain the phenomenon. Although there have been a few attempts to summarize and evaluate several of these hypotheses (Conroy and Hocking 1993; Gifford et al. 2000; Pang et al. 2006), the present review aims to provide a critical and more comprehensive evaluation of these hypotheses than has been available previously.

This review focuses on factors at the plant level that may affect Nm. Elevated [CO2] and other global changes are likely to additionally affect Nm by affecting ecosystem processes that influence N availability to plants, but this is beyond the scope of this review; we refer readers to several recent reviews of these topics (Barnard et al. 2005; Luo et al. 2006; Reich et al. 2006).

Dilution Hypotheses

Perhaps the most frequently mentioned hypothesis about the decrease in Nm under elevated [CO2] is that it results from dilution due to accumulation of non-structural carbohydrates (NSC; Table 1, Hypothesis 1). Dilution by plant secondary compounds has also been proposed as a possible mechanism for decreased Nm under elevated CO2 (Table 1, Hypothesis 2). Most authors who have mentioned dilution have not described the phenomenon in detail; it is therefore uncertain precisely what process is envisioned. Here we distinguish two different meanings of “dilution”, which we call biomass dilution and functional dilution.

Table 1.  Hypotheses to account for the decrease in tissue N concentrations under elevated [CO2]. Papers cited discuss, but do not necessarily endorse, the hypotheses under question
 HypothesisSelected references
1Dilution by carbohydratesGifford et al. 2000, Kuehny et al. 1991, Wong 1990
2Dilution by secondary compoundsGifford et al. 2000
3Decreased N uptake due to decreased transpirationVan Vuuren et al. 1997, Del Pozo et al. 2007, McDonald et al. 2002, Conroy and Hocking 1993
4Less efficient root architecturePritchard and Rogers 2000, Berntson and Bazzaz 1996, Lynch and St.Clair 2004
5Altered root uptake capacityBerntson and Bazzaz 1996, BassiriRad et al. 2001
6Altered mycorrhizal statusBassiriRad et al. 2001, Alberton et al. 2005
7Increased N efficiencyStitt and Krapp 1999, Conroy 1992
8Downregulation of photosynthesisGifford et al. 2000, Fangmeier et al. 1999, 2000
9Increased N lossPang et al. 2006
10Ontogenetic drift of N concentrationColeman et al. 1993, Bernacchi et al. 2007

Biomass dilution occurs whenever the increase in total biomass of a plant or organ under elevated CO2 relative to growth under ambient CO2 is greater than the corresponding increase in total N. One tool that may be usefully applied to examining biomass dilution is graphical vector analysis (GVA) (Haase and Rose 1995; Koricheva 1999). GVA is a technique that allows simultaneous comparison of concentration and content of a chemical component (e.g., N or NSC) and the biomass of the plant organ that contains the component (e.g., leaf) in an integrated graph, thus allowing detection of dilution effects. Our methodology for constructing the vector diagram was modified from Haase and Rose (1995). We extracted results from publications that contained data on biomass and on leaf N and NSC contents. We superimposed the vector diagram for N over the one for NSC to compare their changes in concentration and content relative to changes in leaf mass under elevated CO2 (Figure 1).

Figure 1.

Graphical vector analysis of the effect of elevated CO2 on leaf concentrations of N and non-structural carbohydrates (NSC) in four experiments. All data values are percentages relative to the value at ambient CO2, with the large diamond at (100, 100) representing the value of all measurements under ambient CO2. All other data points are for plants under elevated CO2. Diagonal lines indicate relative leaf mass (i.e. leaf mass under elevated CO2 as a percentage of that under ambient CO2). See text for further explanation. Data is taken from: 1, Pettersson and McDonald (1992); 2,3, Runion et al. (1999), −0.5 Mpa, −1.5 Mpa water potential, respectively; 4, Lewis et al. (2002).

The effect of elevated CO2 on N and NSC is determined by the direction and magnitude of each vector (the arrows in Figure 1). A vector pointing downward and toward the right indicates biomass dilution, while one pointing upward and to the right indicates a concentration effect, in which a constituent increases to a greater extent under elevated CO2 than does biomass as a whole. A horizontal vector pointing right indicates a constituent that increases proportionally with increasing biomass. The length of a vector indicates the magnitude of responsiveness to elevated CO2.

Figure 1 shows four distinct patterns of response to elevated CO2 in N and NSC. Pattern 1, for Betula pendula, shows a response in which N is diluted and NSC are concentrated under elevated CO2. Pattern 2, for Pinus plaustris at high soil water potential, shows a response in which NSC are concentrated under elevated CO2, whereas N decreases not only in concentration but in absolute content as well. This suggests that a mechanism in addition to biomass dilution is contributing to decreasing Nm. Pattern 3, for Pinus palustris at low soil water potential, shows dilution of N, but there is very little increase in the concentration of NSC, suggesting that other biomass constituents are largely responsible for the dilution of N. Pattern 4, for Xanthium strumarium, shows dilution of both N and NSC by increased biomass.

These results demonstrate that one important reason of lower N under elevated CO2 is the relatively small increase of its content compared with other components, i.e. biomass dilution. These results also demonstrate that NSC is in some cases an important component of the diluting biomass. In no case, however, does NSC increase sufficiently to entirely account for the dilution of N. This finding seems likely to hold in general. Poorter et al. (1997) analyzed the chemical composition of the leaves of 27 species grown under elevated CO2 across a wide range of experimental protocols. They found that Nm was decreased under elevated CO2 by an average of 17% on a total biomass basis and by 10% on an NSC-free biomass basis. The content of protein (the major nitrogenous component of leaves) declined not only relative to NSC, but relative to all other classes of organic compounds, including structural carbohydrates, soluble phenolics, lipids, organic acids and lignin. Biomass dilution of N under elevated CO2 may therefore be near-ubiquitous, but NSCs are responsible for only a portion of this effect.

Additional insight into the mechanism of dilution can be obtained through the concept of functional dilution of N. This concept is based on the functional balance concept, which envisions tissue N concentrations as dependent on the relative activities of shoots and roots:

image

where fr and fs are the fractions of plant mass in roots and shoots, respectively, and σr and σs are the mass-based specific activities of roots and shoots in acquiring N and photosynthate, respectively (Davidson 1969; Hilbert 1990; BassiriRad et al. 2001). Within this model, there are three possible causes of a decrease in Nm: a shift in allocation of biomass toward shoots, a decrease in specific root activity, or an increase in shoot specific activity. A decrease in Nm due to increased shoot specific activity can be regarded as functional dilution; Nm declines because of the accumulation of additional photosynthate by shoots.

If we can accept photosynthetic rate as the equivalent of shoot specific activity, functional dilution of N under elevated CO2 appears near-ubiquitous, as enhancement of photosynthetic rates for plants grown under elevated CO2 is consistently observed on either a leaf area or leaf mass basis (Curtis and Wang 1998; Norby et al. 1999; Ellsworth et al. 2004; Ainsworth and Long 2005). The question arises of whether, within the functional balance concept, dilution is solely responsible for decreased Nm under elevated CO2, or whether altered allocation and/or decreased specific root activity play a role as well.

Altered allocation does not appear likely to contribute to decreased Nm; to the extent that elevated CO2 affects allocation it appears generally to increase root mass relative to shoot mass (Norby 1994; Poorter and Nagel 2000; Luo et al. 2006). This should tend to increase, rather than decrease, Nm.

Decreased specific root activity appears more likely than altered allocation to play a role in decreasing Nm. Although a number of experiments have measured the effects of elevated CO2 on root uptake kinetics in solution (BassiriRad et al. 2001), data on specific root uptake of N (i.e. uptake per unit root mass or length) under elevated CO2 for intact plants rooted in solid media (i.e. not in hydroponics) are fairly limited. Table 2 shows the results of all such experiments that we have identified. The consistent trend toward decreased specific uptake is quite remarkable (although the trends were often not statistically significant within individual experiments), as is the overall mean decrease in uptake rate of 16.4%. This result is in striking contrast to the effects of elevated CO2 on uptake of N from solution. Reviews of the literature on short-term (min to h) uptake of NH4+ and NO3 from solution have concluded that results from individual studies are highly variable, with uptake variously increased, decreased or unchanged under elevated CO2 (Luo et al. 1998; BassiriRad et al. 2001). Studies that have measured longer-term uptake of N (days to weeks) in plants grown in hydroponics have also shown variable results, with increases in specific uptake of N under elevated CO2 appearing more common than decreases (Chu et al. 1992; Roumet et al. 1996; Gloser et al. 2002). It therefore appears that the factors that lead to decreased specific N uptake from soils under elevated CO2 are ones not present in hydroponics. This might include aspects of root system architecture and mycorrhizal status or factors affecting the movement of N through soils to plant roots (see the following section for an extended discussion). This suggestion is also consistent with the finding that decreases in Nm under elevated CO2 are larger in soils than in hydroponics (Poorter et al. 1997). In any case, decreased rates of root specific uptake of N appear likely to play a role in the decreased Nm of plants grown in soil.

Table 2.  Effect of growth under elevated CO2 on root specific uptake rates (i.e. uptake per gram root) in studies carried out on intact plants rooted in solid media
PaperSpeciesConditionsMethod and duration of N-uptake measurementsRoot specific uptake rate under elevated CO2 as percentage of that under ambient CO2
  1. When a time-series was reported, measurements are those from the period of greatest uptake rate. Sequential harvest techniques estimate uptake by integrating root mass (or area) and plant N content over the interval between two harvests at which these are measured. The single harvest technique uses a single harvest to obtain the root mass (or area) and N measurements at the end of the experiment. As Israel et al. (1990) point out, this is better thought of as an index of relative N uptake rather than a true estimate of specific uptake. aUptake per root surface area rather than weight.

Coleman and Bazzaz (1992)Abutilon theophrasti28 °CSequential harvests (20 d)59.6%
Coleman and Bazzaz (1992)Abutilon theophrasti38 °CSequential harvests (20 d)60.6%
Coleman and Bazzaz (1992)Amaranthus retroflexus28 °CSequential harvests (20 d)56.6%
Coleman and Bazzaz (1992)Amaranthus retroflexus38 °CSequential harvests (20 d)76.7%
Israel et al. (1990)Glycine max (non-nodulated)High N, high PSingle harvest (27 d)89.2%
Israel et al. (1990)G. max (non-nodulated)High N, low PSingle harvest (27 d)103.1% 
Israel et al. (1990)G. max (non-nodulated)Low N, high PSingle harvest (27 d)87.2%
Rogers et al. (1992)aG. max (non-nodulated) Single harvest (18 d)90.7%
Zerihun et al. (2000)Helianthus annuusLow NSequential harvests (20 d)71.4%
Zerihun et al. (2000)H. annuusMedium NSequential harvests (20 d)100.0% 
Zerihun et al. (2000)H. annuusHigh NSequential harvests (20 d)76.9%
Larigauderie et al. (1994)Pinus taedaLow NSequential harvests (117 d)79.2%
Larigauderie et al. (1994)P. taedaHigh NSequential harvests (117 d)122.9% 
BassiriRad et al. (1996)P. taeda Uptake of 15NH4+(48 h)82.6%
BassiriRad et al. (1996)P. ponderosa Uptake of 15NH4+(48 h)79.3%
McDonald et al. (2002)Populus deltoides Uptake of 15NO3 (7 d)86.4%
Vivin et al. (1996)Quercus robur Uptake of 15NO3 (12 h)100.0% 
  Mean (across all studies) 83.6%
Median (across all studies) 82.6%

Another line of evidence that suggests that dilution is not solely responsible for decreasing Nm under elevated CO2 is the pattern of response of mineral elements other than N. With either functional dilution or biomass dilution, all elements other than the elements assimilated through photosynthesis (C, H and O) should be diluted equally (Loladze 2002). Figure 2 shows that significant decreases are seen in elevated CO2 for four of five macronutrient elements, but that there are significant differences among the elements in the effect of CO2. In particular, N is affected by CO2 more than P and K are. This strongly suggests that the effects of elevated CO2 on mineral element concentrations are mediated through factors involved in uptake and/or metabolism of these elements, not simply through dilution.

Figure 2.

Effect of elevated CO2 on macronutrient concentrations in the edible portions of food crops of studies that measured all elements in the same tissues (n = 42). Results are from an unpublished meta-analysis of the relevant research literature (D. Taub and X. Wang). Geometric means and bootstrapped 95% confidence limits (Adams et al. 1997). Vertical dashed line indicates no effect of elevated CO2.

Hypotheses of Decreased Nitrogen Uptake (Source Effects)

A variety of hypotheses have been advanced that propose mechanisms by which growth at elevated [CO2] might decrease acquisition of N. These mechanisms can be divided into those that affect N uptake by affecting the ability of the soil-root system to supply N (source effects), and those that affect demand for N by the plant, with subsequent effects on N uptake (demand effects). In this section we discuss source-driven effects on N acquisition, with demand-driven mechanisms reserved for the following section.

One of the most consistent effects of growth at elevated CO2 is a decrease in stomatal conductance (Wullschleger et al. 2002; Ainsworth and Rogers 2007), leading to decreased transpiration. Several authors have suggested that this may decrease uptake of those nutrients for which mass flow through soil plays a major role in uptake, including N (as nitrate), Ca, Mg and S (Table 1; Line 3; Barber 1984; Marschner 1995).

In support of this hypothesis, several studies have found elevated CO2 to have similar effects on both transpiration and plant N. Del Pozo et al. (2007), manipulating both N supply and [CO2] in wheat, found transpiration and leaf N (both on a leaf area basis) to be positively correlated. Polley et al. (1999) found that elevated CO2 decreased both whole-plant transpiration and whole-plant N accretion in two C3 perennials.

In either of these studies it is not clear whether transpiration has led to decreased plant N, or whether decreased photosynthetic capacity (manifested by decreased leaf N) has led to decreased stomatal conductance and transpiration (Drake et al. 1997). This caveat does not hold for a study by McDonald et al. (2002), who measured the effect of elevated CO2 on N uptake rates. They found that over a 7-d period, transpiration per gram of root in Populus deltoides was both decreased by 20.4% under elevated CO2 and positively correlated with N uptake per gram of root.

Additional support for a relationship between elevated CO2-mediated decreases in transpiration and in Nm can be found by comparing the responses to elevated CO2 of various soil-derived elements. Figure 2 shows that decreases in concentration under elevated CO2 are largest for those macronutrients that are supplied to roots by transpiration-driven mass flow (N, Mg and Ca) and least for those most dependent on diffusion through soil (P and K). This suggests two possible interpretations. Biomass dilution may decrease the concentrations of all soil-derived elements, while mobile elements are additionally decreased due to restricted mass flow. Alternatively, dilution may decrease the concentration of all soil-derived elements, but this may be partially ameliorated for diffusion-limited elements by increased soil water content allowing more rapid diffusion (Van Vuuren et al. 1997).

Several authors have suggested that growth at elevated CO2 may lead to root system architectures that are less efficient at taking up nutrients, including N (Table 1; Hypothesis 4). Pritchard and Rogers (2000) presented evidence that under elevated CO2, root systems often exhibit increased growth of lateral versus primary roots, leading to shallower rooting, and suggested that this would lead to decreased efficiency of nutrient uptake. Berntson (1994), modeling root systems of Senecio vulgaris, found that changes in root system architecture under elevated CO2 led to decreased efficiency of soil exploitation. Whether this finding would hold for other species and under different growth conditions is unknown.

Another mechanism that might potentially affect N uptake rates is altered uptake capacity of individual roots (Table 1; hypothesis 5). The expectation of most authors has been that elevated CO2 would increase rather than decrease uptake capacity, by providing additional photosynthate to roots for the energy demanding processes of N uptake (BassiriRad et al. 1996, 2001). Studies of root uptake of NH4+ and NO3 from solution have, however, yielded highly variable results (Luo et al. 1998), with a majority of the experimental observations summarized by BassiriRad et al. (2001) showing decreases in NH4 and NO3 uptake rates under elevated CO2. The mechanisms by which elevated CO2 might lead to such decreases are not known. The large variability seen for the effects of elevated CO2 on root uptake capacity suggest that this factor is not primarily responsible for the decrease in Nm typically seen under elevated CO2, but may contribute to the variation seen in magnitude of the Nm response.

Altered mycorrhizal status under elevated CO2 has also frequently been mentioned as a mechanism that might affect plant nutrient status (Table 1; Hypothesis 6). Growth at elevated CO2 often increases the proportion of roots colonized by mycorrhizal fungi (Treseder 2004). It may also have greater positive effects on the mass of mycorrhizal fungi than of their associated plants (BassiriRad et al. 2001) although see Staddon et al. 2002) argued that much of the research on this question was inconclusive. Elevated CO2 may also increase the growth of extraradical mycelium more than it increases the growth of fungal hyphae in direct association with plant roots (Alberton et al. 2005). These changes collectively could potentially increase nutrient uptake efficiency under elevated CO2 by the fungal-plant system, as fungal hyphae have greater uptake capacity on a mass basis than plant roots (BassiriRad et al. 2001). Alternatively, the greater fungal mass under elevated CO2 might provide a larger sink for minerals, so that less is translocated to plants (BassiriRad et al. 2001; Alberton et al. 2005). BassiriRad et al. (2001) found that, under elevated CO2, mycorrhizal plants typically had higher Nm than non-mycorrhizal plants. This suggests that effects on mycorrhizae are unlikely to be responsible for the decrease in Nm typically found under elevated CO2. Differences in the extent of mycorrhizal development may, however, explain some of the variability found in the effect of elevated CO2 on Nm, with colonization by mycorrhizal fungi partially ameliorating the decreases in Nm that are typically seen.

Hypotheses of Decreased Nitrogen Demand

Plants can sustain growth at a lower Nm under elevated than under ambient CO2, as reflected in a lower critical foliar N concentration (the concentration at which biomass production is 90% of maximum; Conroy 1992) and increased plant nitrogen use efficiency (Stitt and Krapp 1999). Assuming that N uptake is at least partially regulated by demand (BassiriRad et al. 2001), this decreased N requirement can lead to decreased whole-plant Nm (Table 1, Hypothesis 7).

Increased plant N efficiency for biomass production under elevated CO2 is almost certainly the product of an increased photosynthetic N use efficiency (PNUE). This increased PNUE appears to be a result both of the effect of [CO2] on the efficiency of carboxylation of RUBP and of decreased investment in photosynthetic and photorespiratory enzymes (Davey et al. 1999; Stitt and Krapp 1999; Gifford et al. 2000).

While there have been suggestions that decreased accumulation of photosynthetic enzymes under elevated CO2 result from a general decline in leaf N status (Stitt and Krapp 1999), there is also evidence for a specific downregulation of photosynthetic enzymes mediated by a sugar-sensing mechanism (Moore et al. 1999).

Photosynthetic enzymes, particularly RUBISCO, make up a large fraction of total leaf N in C3 species (Evans and Seemann 1989). A number of authors have therefore suggested that downregulation of photosynthesis may be partially or largely responsible for the effects of elevated CO2 on Nm in leaves and other photosynthetic tissues (Table 1, Hypothesis 8). Fangmeier et al. (1999, 2000, 2002) have pointed out that downregulation of photosynthetic enzymes in leaves may also decrease N availability to organs such as seeds and tubers that obtain N translocated from catabolized proteins in leaves. Increased PNUE and decreased demand for photosynthesis may therefore explain, at least in part, decreased Nm in a variety of plant tissues.

Elevated CO2-mediated Nitrogen Loss Hypothesis

Although many authors have considered possible CO2 effects on N acquisition, only one study that we are aware of has considered the possibility that elevated CO2 decreases Nm by affecting the rate of plant loss of N (Table 1; Hypothesis 9). Pang et al. (2006) examined N relations in rice grown in pots of nutrient solution under free-air carbon dioxide enrichment. They documented that N loss per pot was greater under elevated than ambient CO2 at both high and low N supply. They attributed this N loss to volatilization of NH3 from senescing plant tissues, and possibly to root exudation of organic N, and suggested that these increases in N loss were responsible for the observed decreases in Nm.

Closer examination of their data suggests that other mechanisms must have also been operating. In their experiment, whole-plant Nm (recalculated from their data) was decreased under elevated CO2 by 26.4% and 26.5% at low and high N, respectively. Had all of the N lost from pots been retained in the plant tissues, Nm would still have been decreased under elevated CO2 by 21.4 and 19.1% at low N and high N, respectively. Effects of elevated CO2 on N loss therefore may have played a role in the observed decrease in Nm under elevated CO2, but other mechanisms were probably more important. Whether the effect of elevated CO2 on N loss is particular to this system also appears to be unknown. Loss of N from senescing leaves does not appear likely to explain the decreases in Nm seen in many experiments in young, non-senescent tissues.

Hypothesis of Ontogenetic Drift in N Concentration

Coleman et al. (1993) proposed that the observed effects of elevated CO2 on Nm may be ontogenetic rather than functional. Plants grow more quickly under elevated than ambient CO2. Comparisons between elevated and ambient CO2-grown plants of the same age are therefore comparisons between plants of different sizes. For seedlings, Nm often decreases during early ontogeny. Differences in Nm between elevated and ambient-CO2 grown plants may therefore simply reflect the fact that the plants are of different sizes rather than any specific effect of CO2 on plant physiology.

Coleman et al. (1993) presented data showing that for two annual species, seedlings grown under elevated CO2 had lower Nm than ambient-grown seedlings when compared at identical ages. When the comparison was made between elevated and ambient-grown plants at the same size, the difference disappeared.

Subsequent studies have in general not fully replicated this finding. The effect of elevated CO2 in some studies is less when plants are compared at the same size rather than the same age, but allometric analyses typically show differences in Nm between elevated and ambient grown plants that are independent of plant size (Lutze and Gifford 1998; Harmens et al. 2001; Marriott et al. 2001; Bernacchi et al. 2007). Some studies also show an effect of CO2 on Nm, with little or no evidence for ontogenetic drift in Nm (den Hertog et al. 1996; Lutze and Gifford 1998; Marriott et al. 2001).

Ontogenetic drift also appears unlikely to explain elevated CO2-mediated differences in Nm seen in perennial plants after more than one season of growth (e.g. Curtis and Wang 1998, Ellsworth et al. 2004) or between seeds compared at full maturity (e.g. Jablonski et al. 2002; Taub et al. 2008). Overall, ontogenetic drift in Nm appears likely to play only a minor role in creating the general pattern of decreased Nm under elevated CO2.

Conclusions

A variety of physiological effects of elevated CO2 on plants have been proposed that could potentially influence plant tissue concentrations of N. It is possible that many or even all of these mechanisms genuinely operate in plants, at least for particular species or under particular environmental conditions. Some mechanisms, such as increases in mycorrhizae or increased capacity for N uptake at root surfaces, appear likely to affect Nm in a positive direction. Other mechanisms, such as biomass dilution, decreased transpiration, decreased efficiency of root architecture and increased N loss are likely to lead to decreased Nm.

A substantial number of literature surveys are in agreement that the mean effect of elevated CO2 on Nm is a 10–15% decrease (Cotrufo et al. 1998; Curtis and Wang 1998; Norby et al. 1999; Jablonski et al. 2002; Ainsworth and Long 2005; Taub et al. 2008). This suggests that physiological changes leading to decreased Nm under elevated CO2 predominate in their effects over factors that would tend to increase Nm.

We suggest that the predominant mechanisms by which elevated CO2 affects Nm are dilution of N in plant tissues by increased concentrations of compounds derived from photosynthate, and decreases in root specific N uptake. Decreased specific uptake is likely due both to decreased demand by the plant (due to increased photosynthetic nitrogen use efficiency and to downregulation of photosynthetic enzymes) and to decreased ability of the soil-root system to supply N. Although a number of mechanisms have been proposed that might account for this decreased N supply, we suggest that decreased transpiration-driven mass flow of N is the mechanism that is best supported by the available evidence.

Two additional mechanisms appear particularly worthy of additional investigation. Both decreased efficiency of root system architecture (Berntson 1994) and increased loss of N by plants under elevated CO2 (Pang et al. 2006) have received experimental support as possible mechanisms contributing to decreased Nm. However, in each case this is based (to the best of our knowledge) on a single study, and additional studies are clearly needed if these mechanisms are to be substantiated.

(Handling editor: Scott Alan Heckathorn)

Acknowledgements

We thank Lisa Anderson for helping in obtaining literature. Jim Coleman and Kelly McConnaughay provided helpful discussion of their research, and the reviewers and editor provided helpful suggestions on the manuscript.

Ancillary