Interactions between atmospheric CO2 concentration and phosphorus nutrition on the formation of proteoid roots in white lupin (Lupinus albus L.)


Rowan Sage. Fax: + 1416 978 5878; e-mail:


Atmospheric [CO2] affects photosynthesis and therefore should affect the supply of carbon to roots. To evaluate interactions between carbon supply and nutrient acquisition, the [CO2] effects on root growth, proteoid root formation and phosphorus (P) uptake capacity were studied in white lupin (Lupinus albus L.) grown hydroponically at 200, 410 and 750 µmol mol−1 CO2, under sufficient (0·25 mm P) and deficient (0·69 µm P) phosphorus. Plant size increased with increasing [CO2] only at high P. Both P deficiency and increasing [CO2] increased the production of proteoid clusters; the increase in response to increased [CO2] was proportionally greater from low to ambient [CO2] than from ambient to high. The activity of phosphoenol pyruvate carboxylase in the proteoid root, the exudation of organic acids from the roots, and the specific uptake of P increased with P deficiency, but were unaffected by [CO2]. Increasing [CO2] from Pleistocene levels to those predicted for the next century increased plant size and allocation to proteoid roots, but did not change the specific P uptake capacity per unit root mass. Hence, rising [CO2] should promote nutrient uptake by allowing lupins to mine greater volumes of soil.


The acquisition of mineral nutrients is dependent on the carbohydrate status of the plant. Root initiation, growth and maintenance, and specific nutrient uptake all require a supply of fixed carbon (Rogers et al. 1999). It is estimated that 10 to 50% of assimilated carbon is respired by the roots, with greater fractions being required during nutrient shortage (Van der Werf, Welschen & Lambers 1992). Nitrogen uptake from the soil can use up to 13% of total photosynthate, mycorrhizae up to 20%, and symbiotic nitrogen fixation 25% (Van der Werf et al. 1992). In addition to their role as substrates in root metabolism, carbohydrates serve as signals in the control of root and shoot development (Roitsch 1999). High sugar concentrations in plant tissues increase lateral root production, and several hormones control root development interactively with carbohydrates (Laby et al. 2000).

Atmospheric [CO2] affects photosynthesis and thus may alter carbon supply to the roots (BassiriRad, Gutschick & Lussenhop 2001). Nutrient supply is closely linked to the ability of plants to respond to rising [CO2], and an important issue is whether elevated [CO2] will enhance the rate at which plants can acquire mineral nutrients (Rogers et al. 1999). At present, [CO2] is rising and could double over this century (Watson et al. 1990). In addition, the current [CO2] is higher than that found over most of earth's recent history, as revealed by ice core data (Petit et al. 1999). During most of the late-Pleistocene epoch, atmospheric [CO2] was below 250 µmol mol−1, with values of 180 µmol mol−1 reported 20 000 years ago (Sage & Coleman 2001). There are a number of ways that variation in atmospheric [CO2] affects nutrient acquisition. Increasing [CO2] may enhance the size of root systems, alter root exudation and stimulate phosphatase activity in the rhizosphere (Prior et al. 1994; Barrett, Richardson & Gifford 1998). Larger root systems can form more infection sites with both nodulating bacteria (Serraj, Sinclair & Allen 1998) and mycorrhizal fungi (Sanders et al. 1998). High [CO2] also increases the number of proteoid roots formed by white lupin (Lupinus albus) under P deficiency (Watt & Evans 1999a). Proteoid roots are dense clusters of determinate rootlets that form on a lateral root in response to P deficiency (Johnson, Allen & Vance 1994). Increased production of proteoid roots enhances surface area exposure of roots to soil particles, and thus promotes acquisition of immobile nutrients.

Most soil P is insoluble and thus available only over very short distances from the root (Gardner, Parbury & Barber 1982). Plants compensate for this in a variety of ways, all of which involve the use of photosynthate. Both increased root growth and proteoid root formation allow access to new regions of the soil (Gahoonia & Nielsen 1998). Carbohydrates are important for the synthesis of ATP required by membrane transporters. Carbon-rich compounds exuded from roots may attract beneficial soil organisms or directly mobilize soil P (Lamont 1982). The exudation of carboxylates, such as citrate and malate, promotes solubilization and uptake of P (Johnson, Vance & Allen 1996b). Increased synthesis of organic acids in the root is accomplished in part by an increase in the activity of phosphoenol pyruvate carboxylase (PEPC), which is induced by P deficiency (Johnson et al. 1996a).

White lupin is an annual legume that is frequently used to study the physiology of proteoid roots (Gardner et al. 1982; Johnson et al. 1996a, b; Watt & Evans 1999a). It grows well on nutrient-poor soils, particularly under low P, where other nitrogen-fixing plants often cannot grow because of the high P requirements of N2 fixation. Both root proliferation and organic acid production may be costly in terms of carbon investment, and plants using these strategies might benefit from higher [CO2] that provides more carbon for investment into nutrient acquisition.

In this report, the interactions between CO2 and P supply on proteoid root production are examined on a whole plant and biochemical level. We specifically examine whether low C availability reduces the ability of plants to acquire limiting amounts of P. This could occur by C deficiency reducing root production relative to shoot growth, reducing P uptake capacity per root segment, or affecting the production of carboxylate chelators such as citrate and malate. Proteoid-root forming species such as white lupin are ideal for studying carbon and nutrient interactions in variable CO2 conditions, because they generally do not form mycorrhizal associations (Gardner, Parbury & Barber 1981). Consequently, proteoid root growth and function can be readily assessed using a hydroponic system.

Materials and methods

Plant material and treatments

Seeds of white lupin (Lupinus albus L. var. Ultra) were germinated in aquarium gravel. At 8 d, they were transferred to hydroponic tubs containing 32 L of 1/8 × modified Hoagland's solutions with sufficient or low P. Solution with sufficient P had 7·5 mm KNO3, 0·31 mm Ca(NO3)2, 0·25 mm NH4H2PO4, 0·125 mm MgSO4, 12·5 µm FeEDTA, 0·69 µm NaH2PO4, 0·012 µm CoCl2·H2O, 3·12 µm H3BO3, 0·25 µm MnSO4·H2O, 0·25 µm ZnSO4·7H2O, 0·062 µm CuSO4·H2O, and 0·062 µm Na2MoO4·2H2O. In the solution low in P, the 0·25 mm NH4H2PO4 was replaced with 0·25 mm NH4NO3, giving a total of 0·69 µm H2PO4. Solutions were changed every 5 d. Phosphorus deficiency (chlorosis and necrosis of the oldest leaves) was evident and proteoid roots were induced by the low P solutions.

Plants were grown in artificially lit controlled environment chambers (E-15; Conviron, Winnipeg, Canada) at 19/15 °C day/night air temperature, with humidity between 50 and 80%, and photosynthetic photon flux density of 440 ± 40 µmol m−2 s−1 for a 16 h photoperiod. Three [CO2] treatments were used: low (200 ± 5 µmol mol−1), ambient (410 ± 20 µmol mol−1, in Toronto), and high (740 ± 40 µmol mol−1). [CO2] was maintained by a CO2-controlling infrared gas analyser (Model WMA-2; PP Systems, Haverhill, MA, USA), which regulated a solenoid valve on a compressed CO2 cylinder for high [CO2], or the flow of air through soda-lime canisters for low [CO2]. Three chambers were used in each replicate experiment, one at each [CO2]. CO2 treatments were rotated between the chambers for each successive experiment. Both P treatments were present in each of the chambers.

Plant harvest and analysis

Plants were harvested 2 weeks after transfer to hydroponics. Plants were either frozen in liquid nitrogen and stored at −80 °C until analysis of carbohydrates and PEPC activity, or oven dried at 70 °C for 3 d for assay of dry weight and P content. All plants were harvested between 4 and 11 h into the light period. Leaf area was measured using a Li-Cor area meter (Model No. LI-3000; Licor Inc., Lincoln, NB, USA).

Phosphorus content

Phosphorus content was determined following digestion in sulphuric acid. Either 5 mg (high-P-treated plants) or 10 mg (low-P-treated plants) of dried tissue were added to 10 mg of powdered zinc and 450 µL of concentrated sulphuric acid (modified from Shirai & Kawashima 1993). Following heating at 400 °C for 20 min, samples were cooled for 5 min and then 100 mg K2SO4 was added. Samples were then heated at 250 °C until clear and diluted to 15 mL. Phosphorus was measured spectrophotometrically at 820 nm after neutralization with 6 m NaOH using ammonium molybdate (Ames 1966).

PEP carboxylase activity and protein content

PEP carboxylase activity was measured after extraction at 2 °C of 0·2 g of root tissue in 3 mL of extraction buffer [50 mm Hepes (pH 7·6) with 1 mm EDTA, 4·9 mm MgCl2, 5 mm dithiothreitol, 1% polyvinylpolypyrolidone, 5 mm amino-n-capriotic acid and 1 mm benzamide] using a Ten-broeck tissue grinder (Wheaton, #62400-518; VWR, Mississauga, ON, Canada; Sage, Pearcy & Seemann 1987). After centrifuging at 8000 × g for 1 min, the extract was divided into two aliquots. One was used to assay PEPC activity and the other was frozen in liquid nitrogen for protein analysis. PEPC activity was assayed in the following manner: a 100 µL aliquot of extract was added to 400 µL of assay buffer [44·4 mm bicine (pH 8·2), 5 mm dithiothreitol, 4·8 mm PEP, 0·2 mm NADH, 2 units mL−1 malate dehydrogenase, 48 mm glucose-6-phosphate and 5·4 mm NaH14CO3, 1600 dpm nmol−1] at 25 °C. The fixation of H14CO3 was stopped after 60 s with 400 µL of 2 m HCl. Acid-stable radioactivity was determined with a scintillation counter (Beckman LS 6000 IC; Beckman, Fullerton, CA, USA). Soluble protein content was determined using a modified Lowry's assay (Yeang, Yusof & Abdullah 1995). Proteins were first precipitated in acetone to avoid the confounding effects of materials that may have been present in the buffer (Johnson et al. 1994).

Carbohydrates and organic acids

Carbohydrate content of leaves and proteoid clusters was measured enzymatically (Hendrix 1993). Soluble sugars were extracted in 80% ethanol at 80 °C, dried and resuspended in distilled water. Starch in the remaining plant material was digested enzymatically, using both α-amylase and amyloglucosidase as described by Hendrix (1993). Both sugar and starch were then analysed with a glucose kit (Catalogue no. 115-A; Sigma, St. Louis, MO, USA) using a microplate reader (Model MR700; Dynatech Laboratories Inc., Middlesex UK).

On the day of harvest, plants were removed from hydroponic basins, rinsed with distilled water, and placed in 100 mL of the appropriate hydroponic solution for 1·5 h. The high-[CO2]-grown plants had more roots than low-[CO2]-grown plants, but the accumulation of organic acids within 1·5 h was small and was unlikely to result in any feedback reactions during the incubation. Plants were then dried and weighed. Root exudate solutions were stored at −80 °C until analysis (Johnson et al. 1994). Samples were vacuum filtered (0·45 µm) and organic acids were separated by high-performance liquid chromatography (Model LC-250B pump; Perkin Elmer, Wellesley, MA, USA) using an Ion Pac ICE-AS6250 × 9 mm column (Dionex, Sunnyvale, CA, USA). The eluent was 0·4 mm HCl with a flow rate of 1·2 mL min−1. Acids were detected with a conductivity detector (550Alltech; Alltech, Nicholasville, KY, USA; 35 °C) without suppression. An autosampler (Waters 712 WISP; Waters, Milford, MA, USA) was used for sample injection.

Phosphorus uptake

The rate of P uptake was determined using excised proteoid root clusters. Only low (200 µmol mol−1) and high (750 µmol mol−1) CO2 treatments were used. Each cluster was equilibrated in a 0·5 mm CaSO4 solution at pH 5·5 for 1 h to remove cold phosphate and stabilize the membranes (Chapin & Van Cleve 1989). Roots were then transferred to uptake solutions for 15 min; all solutions contained 0·5 mm CaSO4 and were adjusted to pH 5·5. Three P-uptake levels were used for low P plants: 1, 5 and 50 µm Na2HPO4. Because of the limited number of clusters on high-P-grown plants, these were only measured at 50 µm Na2HPO4. Each solution contained 5 kBq 32P ml−1 (as orthophosphate; Amersham, Piscataway, NJ). Three clusters were harvested per plant for each of the P levels, and a total of six plants were used for each treatment. Roots were then rinsed three times with a 2·5 mm solution of cold Na2HPO4. Clusters were oven-dried and weighed, then counted intact by liquid scintillation.

Statistical analysis

All data were analysed using a two-way anova for a randomized design in Sigma Stat Version 2·03 (SPSS Inc., Chicago, IL, USA), with P = 0·05 as the critical level of significance. Means were differentiated using a Tukey test. The hydroponic experiments were performed in triplicate, with six samples per replicate; analysis was performed on the means of the six samples.


Growth and allocation

Plant dry weight more than doubled when the [CO2] was increased from 200 to 750 µmol mol−1 at high P (Fig. 1a). At low P, [CO2] had no significant effect on plant size. Leaf area in all of the low-P-treated plants was similar to that of the high-P-treated plants grown at low (200 µmol mol−1) [CO2] (Fig. 1b); however, in the high P treatments, the ambient (410 µmol mol−1) and high (750 µmol mol−1) [CO2]-grown plants both had over 50% more leaf area than those from low [CO2] (P < 0·05), although they were not statistically different from each other. Root dry weight responded to [CO2] and P in the same manner as total plant dry weight; it more than doubled when the [CO2] was increased from 200 to 750 µmol mol−1 at high P but was statistically unaffected by [CO2] in the low-P-treated plants (Fig. 1c). The average length of lateral roots also increased significantly with increasing [CO2][high (750 µmol mol−1) [CO2]: 12·4 ± 0·4 cm, low (200 µmol mol−1) [CO2]: 9·0 ± 1·1 cm, P < 0·05]. The root : shoot ratio of the lupins was little affected by growth [CO2] (Table 1). The increase in root : shoot ratio at high P was only significant in the high-[CO2]-grown plants, with almost a 50% increase from low to high P.

Figure 1.

Effects of growth [CO2] and P treatment on total plant dry weight (a), leaf area (b) and total root dry weight (c) at 22 d after germination (DAG). Nutrient levels: low P, 0·69 µm PO43– (•); high P, 250 µm PO43– (○). Mean ± SE, n = 3 experiments. Different letters indicate statistically different groups at P < 0·05.

Table 1.  Allocation patterns of biomass, total non-structural carbohydrates (TNC) and phosphorus (P) in Lupinus albus L.
Root : shoot
Proportion of total
plant TNC in roots
Proportion of total
plant P in roots
Total plant P
content (mg plant−1)
  1. Harvests at 22 d after germination (DAG). Nutrient treatments: low P, 0·69 µm PO43; high P, 250 µm PO43. Mean ±SE, n = 3 experiments. Superscripted letters indicate statistically different groups at P < 0·05.

200Low P0·42 ± 0·04a0·31 ± 0·02a0·35 ± 0·02a 1·9 ± 0·2a
200High P0·46 ± 0·06ab0·54 ± 0·00b0·46 ± 0·04a14·3 ± 1·8b
410Low P0·36 ± 0·04a0·17 ± 0·02ac0·39 ± 0·05a 1·6 ± 0·1a
410High P0·48 ± 0·02ab0·47 ± 0·07b0·52 ± 0·03ab20·0 ± 3·3c
750Low P0·36 ± 0·05a0·11 ± 0·02c0·44 ± 0·06a 1·6 ± 0·0a
750High P0·53 ± 0·04b0·21 ± 0·02c0·62 ± 0·03b21·1 ± 3·2c

The low-P-treated plants produced four to six times more proteoid roots than those grown under sufficient P, whether measured as a direct count of clusters (Fig. 2a), dry weight of clusters (Fig. 2b), percentage of either lateral root length (Fig. 2c) or total root dry weight (Fig. 2d). Elevated [CO2] resulted in more proteoid roots in low P plants by all measures compared with subambient [CO2], the difference between the highest and lowest [CO2] treatments ranging from approximately 2·5 to five times (significant in all cases at P < 0·05, except for percentage root length). [CO2] had no effect on allocation to proteoid roots in the high-P treatments.

Figure 2.

Effects of growth [CO2] and P treatment on production of proteoid clusters at 22 DAG. Nutrient treatments: low P, 0·69 µm PO43– (•); high P, 250 µm PO43– (○). Mean ± SE, n = 3 experiments for all but the root length data (shows exp. 3 only). Different letters indicate statistically different groups at P < 0·05.

Carbohydrates and phosphorus

There was no significant effect of P or [CO2] treatments on leaf-soluble sugar content, which averaged 31 ± 4 mg g−1 (DW). By contrast, the leaves of low-P-treated plants contained more than twice as much starch as the high-P-treated plants in the same [CO2] treatment (P < 0·005; Fig. 3), and leaves of high-[CO2]-grown plants contained more than three times as much starch as those grown at low [CO2] (P < 0·005, Fig. 3). The response of leaf total non-structural carbohydrates (TNC) to [CO2] and P was similar to that of starch (P < 0·001 for both [CO2] and P, data not shown). No effect of either [CO2] or P supply was apparent on the concentration of carbohydrates in proteoid roots (average TNC was 61·4 mg g−1).

Figure 3.

Effects of growth [CO2] and P treatment on leaf starch content at 22 DAG. Nutrient treatments: low P, 0·69 µm PO43– (•); high P, 250 µm PO43– (○). Mean ± SE, n = 2 experiments. Different letters indicate statistically different groups at P < 0·05.

Total plant P content was unaffected by [CO2] treatment in the low-P-grown plants (Table 1). In the high-P-treated plants, total P content was an order of magnitude higher than under low P (P < 0·001) and did respond positively to increased [CO2] (Table 1). The high-[CO2] (750 µmol mol−1)-grown plants contained 48% more total P than those grown at low [CO2] (200 µmol mol−1) because of larger plant size. In both P treatments, the concentration of P in leaves was 30 to 60% less than in either non-proteoid or proteoid roots (P < 0·05); no significant difference existed between proteoid and non-proteoid roots in terms of [P] (Fig. 4). The effect of [CO2] on [P] was significant only in the low-P-treated plants. Leaves of low-[CO2]-grown plants had more than twice the P concentration of leaves of either the ambient or high-[CO2]-grown plants (P < 0·01), whereas proteoid roots produced at low [CO2] had 30% higher [P] than proteoid roots from plants grown at higher [CO2]. The contribution of additional clusters to P acquisition could not be measured in this experiment because it was impossible to supply insoluble P to proteoid roots in the hydroponic solution. The uptake of P from the nutrient solution was very low in the plants grown in the low-P solution, and the most important source of P for these plants was the reserve in the seed. This is why all low-P-treated plants had similar total P contents at the end of the experiment and why the smaller low-[CO2]-grown plants had higher [P] in their tissues.

Figure 4.

Effects of growth [CO2] and P treatment on tissue P content in (a) low P (0·69 µm PO43–) and (b) high P (250 µm PO43–)-treated plants at 22 DAG. Plant tissue: leaf (•), normal root (○), and proteoid root (▾). Mean ± SE, n = 3 experiments. Different letters indicate statistically different groups at P < 0·05.

In the high-P-treated plants, there was a doubling of the proportion of the TNC allocated to the roots; the high [CO2] treatment decreased this proportion almost three-fold relative to low-[CO2]-grown plants (Table 1). [CO2] treatment had no significant effect on the proportion of plant P allocated to roots in the low P treatment, but the fraction of P present in the roots of the high-P-treated plants increased 36% from low to high [CO2]. Within the high [CO2] treatment, the high-P-treated plants thus allocated significantly more P to roots than P-deficient plants.

Proteoid root activity

The total activity of PEPC extracted from the roots of the low-P-treated plants was approximately twice as high as that from the high-P-treated plants (P < 0·001) regardless of whether it was measured as a function of root fresh weight or soluble root protein. Growth [CO2] had no effect on total PEPC activity in either the high or low P treatments (low-P-treated plant 0·13 ± 0·00 nmol CO2 g−1 FW s−1, high-P-treated plants 0·06 ± 0·01 nmol CO2 g−1 FW s−1). Total protein content did not differ significantly with P treatment (data not shown).

Low-P-treated plants produced more citrate than did high-P-grown plants. On a whole plant level, increasing [CO2] increased the citrate exudation of low-P-grown plants (Fig. 5a), but this [CO2] effect was eliminated when expressed per unit proteoid cluster mass (Fig. 5c). Malate responded differently than citrate to both P and CO2 supply; the increase in malate exudation under low growth P was significant only on a whole plant basis at low and ambient [CO2], and malate exudation was enhanced in P-sufficient plants under high [CO2]. Malate exudation was undetectable in low and ambient [CO2] treatments but, at high [CO2], it was similar to low-P-treated plants (Fig. 5b & d).

Figure 5.

Effects of growth [CO2] and P treatment on root organic acid exudate at 22 DAG. Nutrient treatments: low P, 0·69 µm PO43– (•); high P, 250 µm PO43– (○). Mean ± SE. Different letters indicate statistically different groups at P < 0·05.

The specific P uptake of excised proteoid root segments was unaffected by growth [CO2] (Fig. 6). When data from all [CO2] treatments were pooled, low-P-grown plants (uptake 3·3 ± 0·2 µmol P g−1 root h−1) showed significantly faster P uptake rates than high-P-grown plants (uptake 2·5 ± 0·2 µmol P g−1 root h−1) at 50 µm PO4 in the external solution (P < 0·01).

Figure 6.

Effects of growth [CO2] and P treatment on specific uptake of P at 22 DAG. Nutrient treatments: low P, 0·69 µm PO43– (filled symbols); high P, 250 µm PO43– (empty symbols). [CO2]: 750 µmol mol−1 (▴) and 200 µmol mol−1 (•). Mean ± SE, n = 2 experiments. Different letters indicate statistically different groups at P < 0·05.


In the time between the maximum of the last ice age 20 000 years ago and the future peak of the anthropogenic carbon load in the atmosphere, [CO2] will likely increase three- to four-fold (Watson et al. 1990). Because nutrient limitations are more pronounced at high [CO2] in the absence of nutrient additions, the ability of plants to use the extra carbon to improve nutrient acquisition will influence their ability to exploit potential benefits conferred by CO2 enrichment (BassiriRad et al. 2001). Additional carbon may enhance specific uptake capacity by promoting the activity of membrane transporters and the exudation of compounds such as chelators; alternatively, plants may simply produce more roots, with the activity of the root segments remaining constant. In white lupin, the imposition of P deficiency alone results in a pattern of increased production of proteoid roots, higher exudation of organic acids, and enhanced Vmax for P uptake by root segments (Johnson et al. 1996b). Increases in carbon supply also stimulate proteoid root proliferation, but in contrast to the effect of P deficiency, the P uptake capacity of root segments and the production of organic acids are not affected. At 200 µmol mol−1 CO2, photosynthesis is impaired 30 to 50% relative to current [CO2] (Sage & Reid 1992; Tissue et al. 1995; Sage 1995) and thus carbon skeletons could be in short supply. Increased organic acid production occurs as a result of increased PEPC activity in P-deficient roots (Johnson et al. 1996a). CO2 deficiency does not affect PEPC capacity, nor does it affect the supply of carbon skeletons to PEPC, as indicated by the ability of root segments to exude constant levels of citrate and malate (Fig. 5; Watt & Evans 1999a). This stability in carbon supply to the roots likely reflects the proportional decline in bulk root and shoot mass with [CO2] reduction, as indicated by our root : shoot data.

White lupin plants grown at high [CO2] produce more than twice as many proteoid roots as plants grown at the much lower Pleistocene levels. The largest difference in the proportion of the root system allocated to proteoid roots found in this study is between the plants grown at low and ambient [CO2], indicating that proteoid root allocation in the current atmosphere may be approaching the limit of the lupins’ ability to respond to increasing levels of atmospheric [CO2]. This is similar to the [CO2] response of the C3 annual Abutilon theophrasti (Dippery et al. 1995). In A. theophrasti, increases in root growth in response to CO2 enrichment are substantially greater between 150 and 350 µmol mol−1 than between 350 and 700 µmol mol−1. Notably, nitrogen acquisition declined in A. theophrasti at 150 µmol mol−1, and this contributed to a collapse of photosynthesis and growth capacity (Tissue et al. 1995). At 270 µmol mol−1, however, the synergism between nitrogen acquisition and low [CO2] was not apparent. These results, and our observations in white lupin, indicate that plants modulate root and shoot production so as to prevent carbon starvation in roots down to at least 200 µmol mol−1 CO2. The reduction in nutrient acquisition at 150 µmol mol−1 in A. theophrasti indicates this control cannot be maintained at a [CO2] below that experienced in the late-Pleistocene.

Interestingly, the increase in partitioning to proteoid roots is not matched by an increase in partitioning to the root system in general. Root : shoot ratio is little affected by [CO2] and is slightly increased under high P. This result is unusual in that many C3 species respond to nutrient deprivation and elevated [CO2] with proportional increases in root mass (Stulen & den Hertog 1993). Our results are consistent with previous results from P-deficient lupins (Johnson et al. 1994; Neumann et al. 1999), and they are not complicated by difficulties in recovery that often occur in soil environments. Hence, although atypical with respect to most plants, species that form proteoid roots can respond to nutrient challenges through proteoid root production rather than increases in biomass allocation to the root system as a whole. Relative to coarse root growth, proteoid root production is a more effective means of acquiring immobile nutrients; it is also more efficient in terms of carbon and energy requirements (Gardner et al. 1981). Similarly, responses to nutrient deficiency in species infected with mycorrhizal fungi in combination with high [CO2] often involve enhanced mycorrhizal infection and proliferation, rather than increased root to shoot ratio (BassiriRad et al. 2001).

Any increase in the growth rate of plants at high [CO2] can only be maintained if nutrient uptake is increased in proportion to increased plant growth (BassiriRad et al. 2001). Nutrient uptake capacity is a function of absorptive surface area and specific uptake per root segment. Root surface area reflects total root length and the density of proteoid root clusters, whereas chelator production and membrane transport activity influence specific uptake capacity (Raghothama 1999). Lupins can grow in nutrient-poor soil where most P is insoluble and bound to soil colloids, so uptake capacity is dependent upon the extrusion of protons and chelators from proteoid roots (Gardner et al. 1982). In lupin, the only effect of [CO2] on P uptake was through changes in the root length and biomass allocated to proteoid roots; specific P uptake and citrate production were unaffected by growth [CO2]. These responses allow the [CO2] effect on P acquisition to be estimated. Plant size in the low-P treatment increased 1·4-fold from Pleistocene to post-industrial [CO2], whereas proteoid root biomass increased 5-fold. Given the lack of a [CO2] effect on specific uptake capacity, the potential increase in P acquisition per unit plant biomass is estimated at 360%. Thus, phosphorus supply to plant tissues could be maintained or even increased at higher [CO2], provided the added root mass does not deplete soil P-levels or compete with adjacent proteoid clusters. At high [CO2], the additional root mass is often accompanied by an increase in the lateral spread of root systems (Prior et al. 1994; Pritchard & Rogers 2000). Increased root extension would reduce interference between adjacent proteoid clusters, but could increase competition between plants (Berntson & Woodward 1992). Under such circumstances, increased proliferation of proteoid clusters could enable proteoid-root species to exploit bound soil P more intensively than non-proteoid species, possibly resulting in greater ecological success.

The lack of a response of proteoid root physiology to variation in [CO2] is similar to the response of nitrogen fixing nodules in that the nitrogen fixation rate in individual nodules often does not change with variation in [CO2] (Diaz 1996; Cabrerizo et al. 2001). The responses differ, however, in that the proportion of the root system allocated to nodules generally does not change (Diaz 1996) whereas the proportion of root systems allocated to proteoid clusters increases from low to high [CO2]. Mycorrhizae, the means by which most plants obtain soil P, show a variable response; the number of infection sites formed under high [CO2] may remain constant, as with nodules (Jongen, Fay & Jones 1996), or increase, as with proteoid clusters (Rillig, Field & Allen 1999). Exposure of host plants to high [CO2] is beneficial to mycorrhizae, increasing hyphae length and density in the soil (Rillig et al. 1999); however, the increase in carbon allocated to mycorrhizae does not always increase the proportion of P transferred from the fungus to the host plant (Sanders et al. 1998). In some cases, mycorrhizae become parasitic at high [CO2], in that P transfer to roots declines as carbon supply increases (BassiriRad et al. 2001). Here, we saw no evidence for increased competition for P by roots grown in high [CO2], because the proportion of P in roots did not increase with P deficiency.

The mechanism stimulating the formation of proteoid roots under P deficiency is not known. The signal for proteoid root formation is likely to be shoot derived (Watt & Evans 1999b). It is however, modified to some degree by the soil environment, because proteoid clusters preferentially form in areas of high nutrient concentration (Skene 1998). The signalling pathway in proteoid root formation may act in part through phytohormones. Exogenous auxin stimulates the formation of proteoid roots at high P; interestingly, these clusters do not show increased PEPC activity or root exudation, showing that auxin alone is insufficient for the whole proteoid root response (Gilbert et al. 2000). Carbohydrates can serve as signals in the control of root development and lateral root initiation (Roitsch 1999). For example, the application of exogenous sucrose to one part of a root system promotes localized root growth similar to the proliferation of roots and proteoid clusters found in nutrient rich portions of soil (Bingham, Blackwood & Stevenson 1998). Notably, the effect of carbohydrates on proteoid root formation requires P deficiency, indicating that nutrient status is the master control whereas carbon status is secondary.

Lupins change the relative proportions of organic acids exuded from root systems in response to environmental influences, with plants grown at high [CO2] and high P producing more malate on a whole-plant basis. In lupins, proteoid roots produce more citrate than malate, whereas the reverse is true of non-proteoid roots (Massonneau et al. 2001). The proportion of the root system that is allocated to proteoid clusters determines the relative importance of each organic acid; this explains the large apparent increase in malate exudation in the high-[CO2] and high-P treatment, as these plants were large with large root systems and few proteoid clusters.

Increasing [CO2] increases the size of white lupins and the biomass allocated to proteoid roots. If other plants forming cluster roots follow this pattern, ecosystems that are rich with these species, such as the Fynbos in South Africa, may not experience greater P-deficiency at high [CO2]. Proteoid root proliferation could sustain higher biomass productivity by transferring a greater fraction of the immobile P-pool from the soil to the biosphere. This could have an immediate benefit for neighbouring species; for example, grasses obtain more P from insoluble soils pools when grown with white lupin (Kahm et al. 1999). All species in an ecosystem would benefit as well when the additional P moves through the trophic network and is recovered from decomposing biomass. Conversely, during the low [CO2] episodes of the past, a greater proportion of the bound P may have remained in the soil, causing a synergistic reduction in ecosystem productivity.


We thank Deborah Allan and lab for their kind gift of seeds, Bob Jefferies and Nancy Dengler for comments and advice, and Art Fredeen and David Kubien for helpful comments on the manuscript. This research was supported by grant OGP0154273 from the Natural Sciences and Engineering Research Council of Canada (NSERC) to R.S. and an NSERC scholarship and Ontario Graduate Scholarship in Science and Technology to C.C.