Elevated CO2 increases carbon allocation to the roots of Lolium perenne under free-air CO2 enrichment but not in a controlled environment
Summary
- •
Grass swards often show a higher root–shoot ratio of dry matter (R : SDM) at elevated [CO2] than at ambient [CO2]. However, it is not known whether this is a result of a sustained increase in C allocation to the roots.
- •
The effects of free-air CO2 enrichment (FACE) on carbon allocation to roots in established Lolium perenne (perennial ryegrass) swards during regrowth were investigated by means of 14C labelling and compared with data from a controlled environment experiment.
- •
Elevated [CO2] in the field increased root dry matter (109%), R : SDM (44%) and, in accordance, root–shoot 14C ratios (R : S14C, 39%), as well as the intercept of the allometric line. By contrast, elevated [CO2] did not affect the allometry of the plants under controlled conditions. However, at low N concentration in the nutrient solution, the R : SDM and the intercept of the allometric line increased.
- •
It is suggested that the remarkably different responses to elevated [CO2] under field and controlled conditions result from differences in both N availability for growth, and C sink strength of the shoots.
Introduction
Since the beginning of industrialization, the atmospheric partial pressure of CO2 (pCO2) has been steadily increasing from 27 Pa to its present value of 36 Pa CO2. Models predict that the atmospheric pCO2 will increase to about 70 Pa CO2 in the second half of the twenty-first century (Houghton et al., 1996). Grasslands account for a considerable land area (FAO, 1995) and are considered to have a high CO2 sink capacity (Parton et al., 1995). Thus, grasslands are suggested as playing an increasingly important role in the earth’s carbon cycle.
Various experiments have shown that grass plants exhibit a higher root–shoot ratio (R : SDM) when grown at elevated pCO2 (Farrar & Williams, 1991; Rogers et al., 1996; Soussana et al., 1996; Cotrufo & Gorissen, 1997; Hebeisen et al., 1997; Schapendonk et al., 1997; van Ginkel et al., 1997), due mainly to an increase in root DM. Assessments of the R : SDM allometry of seedlings grown in short-term experiments in controlled environments revealed that ontogenetic drift may be responsible for a shift in R : SDM at elevated pCO2 (Baxter et al., 1997; Lutze & Gifford, 1998; Stirling et al., 1998). However, it was not clear whether older cut plants still show a higher R : SDM at elevated pCO2 during regrowth. At elevated pCO2 under field conditions, Lolium perenne exhibited a strong increase in single-leaf photosynthesis (Rogers et al., 1998; Isopp et al., 2000) but only a weak increase in yield (Hebeisen et al., 1997; Suter et al., 2001). This discrepancy may be explained in part by a greater proportion of fixed C that is allocated to unharvested plant parts. Well-established swards under field or field-like conditions also show a higher R : SDM at elevated pCO2 (Hebeisen et al., 1997; Schapendonk et al., 1997; Suter et al., 2001). It was not clear, however, whether this increase in root DM of field-grown swards was solely a result of: (1) a larger amount of undecomposed (Jongen et al., 1995) dead roots together with an unchanged amount of live roots at elevated pCO2; or (2) an increase in root longevity (Eissenstat et al., 2000; Pritchard & Rogers, 2000), resulting in a slower turnover rate of the root DM pool. In both cases, despite a greater root DM, the C supply to the root fraction could be unaffected by elevated pCO2. Thus, the question arose: How does the allocation of C in (1) field-grown swards and (2) plants in a controlled environment change at elevated pCO2? In order to answer this question, the C allocation in the plant must be determined. Single measurements at the end of a regrowth period or average values of the regrowths in a growing season may mask dynamic processes in C allocation during a particular regrowth period.
In order to elucidate the effect of elevated pCO2 on C allocation in grassland ecosystems, the dynamics of 14C partitioning was examined in Lolium perenne (perennial ryegrass) model swards in the Swiss FACE (free-air carbon dioxide enrichment) facility during a vegetative regrowth period. In a second experiment, the dynamics of root and shoot growth of regrowing spaced plants on a sand culture with nutrient solution were examined in controlled environment chambers to elaborate the effect of contrasting experimental conditions.
Materials and Methods
Field experiment
Experimental site
The field experiment was conducted in the Swiss FACE array (Hebeisen et al., 1997) at Eschikon (8°41′ E, 47°27′ N) near Zurich, 550 m above sea level. The soil, an eutric cambisol, consisted of about 36% sand, 33% silt and 28% clay and was a clay loam. The organic matter content varied from 2.9 to 5.1% and the pH (water extracted) ranged from 6.5 to 7.6. The available P and K (CO2-saturated water extractable) ranged from 1.2 to 6.0 mg P kg−1 soil and from 18 to 47 mg K kg−1 soil, respectively (Lüscher et al., 1998). To maintain nonlimiting levels of phosphorus and potassium in intensive grassland, 4.3 g P m−2 yr−1 and 24.6 g K m−2 yr−1 were applied once per year in spring. The annual amount of N fertilizer was 56 g N m−2, which is markedly higher than the 15–30 g N m−2, typical of grasslands in this region. The portions of N applied during the growing season were adjusted to the expected DM production in each regrowth period, as described by Zanetti et al. (1996): highest in spring and lowest towards the end of the growing season. Nitrogen was applied as NH4NO3 every 4 wks beginning at the start of the growing season. Table 1 lists the meteorological data. During the observation period the temperature was about 0.5°C above and the precipitation about 45% below the 1993–99 average. The sum of global radiation was at about average.
| Date | Temperature (°C) | Precipitation (mm) | Global radiation (MJ m−2) |
|---|---|---|---|
| 30 June–6 July | 16.6 | 14.0 | 120.7 |
| 7–13 July | 17.2 | 2.3 | 111.1 |
| 14–20 July | 19.3 | 8.1 | 154.4 |
| 21–27 July | 21.2 | 32.1 | 130.6 |
| 28 July–3 August | 17.7 | 20.6 | 95.2 |
| 4–10 August | 20.9 | 0.0 | 157.1 |
| 11–17 August | 22.3 | 0.1 | 155.9 |
| 18–24 August | 17.3 | 47.8 | 91.4 |
| 25–31 August | 13.0 | 0.0 | 131.8 |
| Mean | 18.4 | – | – |
| Sum | – | 125.0 | 1148.2 |
Preparation and treatments
A sward of Lolium perenne, which grew under FACE conditions for more than 3 yr, was removed by applying glyphosate [N-(phosphonomethyl)glycine] followed by shallow tillage in mid-July in 1996. Individual plants of Lolium perenne L. cv. Bastion, a widely used forage plant species in temperate regions, were sown in 15-ml pots in soil taken from the FACE area at the beginning of the last week in July and grown in a greenhouse for 6 wks. In the first week of September, the plants were transplanted to the main plots (9.24 × 3.15 m) at a spacing of 7 cm (204 plants per m2). The swards were cut 5 cm above the ground in the first week of November. In the spring of 1997 and 1998, all the plots were weeded by hand before the onset of the growing season.
The experiment was designed as a split-split-plot with three blocks. The blocks were set up according to the crops grown in the year preceding the installation of the FACE facility (4 yr before the start of the experiment described here). These crops were winter wheat (blocks one and three) and grass/white clover mixture (block two). The main plot treatment was pCO2. The levels of pCO2 were ambient (36 Pa) and elevated (60 Pa). The latter was obtained by using the FACE technique, as described by Lewin et al. (1992). Briefly, three fumigated circular areas (18 m diameter) were installed in the spring of 1993. Three nonfumigated control areas of the same size were established at the same time. Fumigation was done by a circular array of 32 vertical vent pipes, which were fed with CO2-enriched air through a 18-m toroidal distribution tube. The valves of the vent pipes were computer controlled. At wind speeds of more than 0.4 m s−1 only the valves of the pipes on the windward side were opened. The CO2-enriched air was, thus, distributed over the experimental area by the wind. At wind speeds below 0.4 m s−1 the valve of every second vent pipe was opened. An infrared gas analyser, connected to a computer, continuously measured the actual pCO2 in the air, which was sampled in the centre of the experimental area, 0.1 m above the canopy. Over the whole experimental period, the 1-min averages of pCO2 were within the target pCO2 of 60 ± 6 Pa for 90–94% of the fumigation time, depending on the ring. The seasonal average of the pCO2 was 60 ± 0.2 Pa. Fumigation was continuous during daylight hours and lasted from spring until November when the air temperature was at least 5°C (2 m above the ground), assuming that plant growth was marginal and that CO2-induced effects on photosynthesis were weak at low temperature (Long, 1994). There was a distance of at least 100 m between fumigated and control areas in order to eliminate the possibility of interactions among the areas, (e.g. ambient areas being affected by elevated pCO2). Each area contained one main plot.
Pretreatments were applied during the establishment of the swards and until the observations started in the summer of 1998 in order to produce plants of different morphology. The pretreatments were combinations of cutting height (4 cm and 8 cm) and cutting frequency (intervals of 4 wks and 8 wks, resulting in eight and four cuts per year, respectively) and were randomly assigned to the four subplots (2.31 × 3.15 m) within each main plot. Four sub-subplots (0.21 × 0.21 m, surrounded by a 0.21-m border) were randomly chosen from each subplot. The sub-subplots were used for destructive harvests and for 14C pulse-labelling.
14 C pulse-labelling
In the established sward in the field, changes in the size of the big pool of roots, consisting of growing and old dying roots, do not represent C allocation to the roots. In order to measure C allocation to the roots, 14C pulse-labelling was carried out. Just before the start of the labelling of the sub-subplot, 1 ml of 9.25 MBq 14C Na2CO3 solution (50 mm Tris, pH 7) and 29 ml distilled H2O were drawn into a syringe. At the start (t0) of labelling, 1 ml of the diluted solution was injected into a 50-ml reaction tube containing 5 ml of lactate (85% in H2O). A pump connected to the reaction tube blew air (10 l min−1) from the surroundings (intake at a distance of about 0.5 m from the sub-subplot) through the solution in the reaction tube and into the Plexiglas labelling cuvette (0.35 × 0.35 × 0.4 m), which was placed over the sub-subplot; the stream of air carried labelled CO2 into the cuvette, which was vented to the environment. A small battery-powered fan inside the cuvette prevented condensation on the surface of the cuvette and mixed the labelled CO2 with the air. The injections were repeated at 1-min intervals. The cuvette was removed 1 min after the last injection (t30). The labelling procedure usually started at 10 : 00 hours and was conducted simultaneously in both CO2 treatments of the same block. The order of labelling of the differently treated plots was randomized per block but was the same in both pCO2 treatments of the respective block. One block was labelled each day. Labelling took place on 20–22 July, 3–5 August, 17–19 August and 31 August to 2 September in 1998.
Data collection and measurements
Measurements were made after 14, 28, 42 and 56 d of regrowth from 22 July to 4 September 1998. The plant material from an area of 0.21 × 0.21 m (sub-subplot) was harvested destructively 46 h after pulse labelling. The shoot material was divided into stubble (below cutting height) and yield (above cutting height), using a stratified clipping technique. The above-ground plant material was harvested using a metal frame to ensure exact cutting height. The roots were sampled to a depth of 10 cm (0.21 × 0.21 m), which accounts for 85–90% of the total root mass of Lolium perenne below the sampling area at the experimental site (Hebeisen, 1997), and the roots were washed carefully. Furthermore, elevated pCO2 did not affect the vertical distribution of the roots (Zimmermann, personal communication). Harvested sub-subplots were replaced by turfs from the same pretreatment to avoid gaps in the subplots. Green and necrotic (visually determined) plant material from each fraction were separated. The plant material was heated in a microwave oven (Popp et al., 1996) to stop any enzymatic activity and dried in a ventilated oven at 65°C for 48 h.
14 C analysis
The plant material was ground in a sample mill (Tecator Cyclotec 1093, Tecator AB, Höganäs, Sweden) and processed further in a ball mill (Retsch MM2, Schieritz und Hauenstein AG, Arlesheim, Switzerland). The powdered sample (22.5 ± 2.5 mg) was put into a 20-ml glass vial (Packard Instrument Company, Meriden, CT 06450, USA) and a suspension of 4 mg cellulase (crude powder from Trichoderma viride) and 4 mg maceroenzyme (pectinase, crude powder from Rhizopus sp.) in 200 µl phosphate buffer (2 mm NaH2PO4, 0.5 mm CaCl2, pH 6) was added. The capped vials were placed on a shaker and incubated at 37°C for 20 h. One millilitre of Soluene-350 (Packard Instrument Company) digesting solution was added to each vial, and the re-capped vials were incubated again at 37°C for 20 h. After incubation and after the samples had reached ambient temperature, 15 ml Hionic-Fluor (Packard Instrument Company) were added to each sample. The vials were capped and shaken to homogenise the digested and dissolved plant material before liquid scintillation counting (Packard 2500TR, Packard Instrument Company). Vials without plant material and enzymes were used to determine the 14C background. Internal standards with a known amount of 14C were used to correct for quenching.
Allometry
Logarithmic shoot DM (sum of stubble and yield) was plotted against logarithmic root DM. The allometric relation was obtained as the linear regression function in the double log plot (Farrar & Williams, 1991; Baxter et al., 1994). This line shows the relative growth rates of C-acquiring tissue (shoot) in relation to the relative growth rates of nutrient-acquiring tissue (root) of the sward and is expressed as
Loge(root DM) = b + k Loge(shoot DM).
Statistical analysis
The statistical analysis was carried out using the GLM procedure of the SAS package (SAS Institute, 1996). The model was a split-split-plot with pCO2 as the main plot factor; thus, pCO2 was tested with the pCO2 × block interaction as the error term. This is a weak test, and pCO2 effects are more readily detected by the dynamics of the response of the pretreatment effects to elevated pCO2 (e.g. pCO2 × cutting frequency × sampling date and pCO2 × cutting height × sampling date). Dry matter data and root–shoot ratios were Loge transformed before the analysis of variance in order to obtain normal distribution and homogeneity of variance. The reported means were obtained by back-transformation. For all the traits examined, the cutting height affected neither the response to CO2, nor the response to cutting frequency. Thus, the data of cutting height were pooled. The linear regression functions for the different pCO2 treatments were compared using Fisher’s t-test (Zar, 1984).
Controlled environment experiment
Plant material and growth conditions
Lolium perenne L. cv. Bastion plants were grown from seed in boxes filled with silica sand (particle size 0.7–1.2 mm) to 5 cm and placed in Conviron PGV 36 controlled environment chambers (Conviron, Winnipeg, Manitoba, Canada) set at 35 or 60 Pa pCO2. The levels of pCO2 were monitored by WMA-2 infrared gas analysers (PP-Systems, Hitchin, Herts, UK). Excess CO2 was removed by forcing air through soda lime pellets (Roth, Karlsruhe, Germany). The daylength was set to 16 h with 12 h at maximum PPFR (500 µmol m−2 s−1 at the pot level). The PPFR was increased and decreased at 2-h intervals at the beginning and end of the day, respectively. Very high-output cool white fluorescent tubes (Osram Sylvania Inc., St Marys, PA, USA) and incandescent bulbs (Osram Sylvania Inc.) were used to obtain the desired PPFR. Relative humidity was maintained at 80%. Air temperature was set to 18°C/13°C (day/night). Plants were irrigated twice a day with a nutrient solution containing 7.5 mm NO3− (Hammer et al., 1978). Iron (0.124 mm EDTA ammonium iron(III)salt; Fluka, Buchs, Switzerland) was supplied without adjusting the pH. Plants and pCO2 treatments were rotated between growth chambers each week.
After 17 d, vigorous plants with one tiller and two to three leaves were selected in each pCO2 treatment, trimmed to 5.5 cm of shoot and 2.5–3 cm of roots and transplanted singly into cylindrical pots 7 cm in diameter and 25 cm high, filled with 1.5 kg of silica sand, as described earlier. The level of elevated pCO2 was increased to 70 Pa. Over the whole experimental period, the average of the low pCO2 was 35 ± 0.8 Pa for 96% of the time and the average of the high pCO2 was 60 ± 0.9 Pa and 70 ± 1.1 Pa (before and after planting in individual pots, respectively) for 95% of the fumigation time. In order to guarantee a controlled nutrient supply, pure silica sand and nutrient solution were chosen to avoid N sequestration/mineralization, which may occur in field soil. Nutrient solution (30 ml) was added to each pot twice per day. The N concentration was changed to the experimental levels 39 d after sowing. Within each of the six replicates, plants were randomly allocated to sampling dates. Nutrient solutions were prepared according to (Hammer et al., 1978), without adjusting the pH, using 0.124 mm EDTA ammonium iron(III)salt (Fluka) as a source of iron. The solution was modified as follows: in the low-N regime, the KNO3 and Ca(NO3)2 were replaced by 0.5 mm KCl, 1 mm K2SO4, 0.5 mm Ca(NO3)2 and 2 mm CaSO4 (final concentration of NO3− was 1 mm); in the high-N regime the concentration of Ca(NO3)2 was changed to 6.25 mm (final concentration of NO3− was 15 mm, a concentration that provides ample N under these experimental conditions). Pots were flushed with adequate amounts of deionized water before the first application of new nutrient solutions (100 ml per pot twice a day). To ensure a nonlimiting water supply with increasing plant size, the amount of nutrient solution was increased to 150 ml per pot (twice a day) 63 d after sowing.
Sequential harvests during regrowth
Forty-five days after sowing (28 d after transplanting into pots), plants were defoliated to 4 cm above the ground. Regrowth was monitored at four harvests, 7, 11, 17 and 28 d after defoliation. For each harvest, one set of plants, consisting of six replicates, was sampled. The sand was carefully removed and the plants separated into roots, stubble (leaf sheaths and enclosed parts of the growing leaf), yield and necrotic tissue. Leaves were classified as necrotic when more than two-thirds of their area had senesced. The dry weight of oven-dried (48 h at 65°C) material was determined. Since the root pool contained young roots only, the relative increase in root weight to shoot weight in these young plants is taken as a measure of C allocation (no 14C labelling).
Statistical analysis
Dry matter and root–shoot ratios were Loge transformed before statistical analysis in order to obtain normal distribution and homogeneity of variance. Means shown were obtained by back-transformation. Statistical analyses were carried out using the GLM procedure of the SAS statistical analysis package (SAS Institute, 1996). The experiment was a complete randomized block design; the factors CO2, N and sampling date were varied, and each treatment was replicated six times. The linear regression functions for the different pCO2 and N levels were compared using a Fisher’s t-test (Zar, 1984).
Results
Dry matter
In the field experiment, averaged over the whole observation period, total DM at elevated pCO2 (977 g m−2) was 65% higher (P < 0.05) than DM at ambient pCO2(Fig. 1a). This CO2-induced difference was mainly due to a 109% increase (P < 0.05) in root DM, from 184 g m−2 at ambient pCO2 to 384 g m−2 at elevated pCO2 (Fig. 1b), and to a 53% increase (P < 0.05) in stubble DM, from 310 g m−2 at ambient pCO2 to 473 g m−2 at elevated pCO2 (Fig. 1c). The yield DM (Fig. 1d) exhibited a 378% increase (P < 0.0001) during regrowth; it showed 30% higher values at elevated pCO2, but this difference was not significant.
(a–d) Development of total DM (a), DM of roots (b), stubble (c) and yield (d) during 56 d of regrowth of 2-yr-old field-grown Lolium perenne swards at ambient (36 Pa, open symbols) and elevated (60 Pa, solid symbols) pCO2 and after infrequent (four cuts per year, circles) and frequent (eight cuts per year, triangles) cutting treatments before the examined regrowth. (e–h) Total DM (e), DM of roots (f), stubble (g) and yield (h) during 28 d of regrowth of spaced Lolium perenne plants grown in a controlled environment on sand culture with nutrient solution containing low N (1 mm NO3−, hexagons) or high N (15 mm NO3−, squares) at ambient (35 Pa, open symbols) and elevated (70 Pa, solid symbols) pCO2. Error bars = SE, n = 6.
In the controlled environment experiment, total plant DM (Fig. 1e) increased (P < 0.0001) by 1100% during regrowth. Averaged over the whole observation period, 54% more DM (P < 0.0001) was measured at elevated pCO2 than at ambient pCO2. The responses of the DM (Fig. 1f–h) of the yield (60%), stubble (43%) and roots (47%) to elevated pCO2 over the observation period were all highly significant (pCO2P < 0.0001). Plants grown under a high-N regime exhibited 130% more (P < 0.0001) total plant DM than low-N plants (Fig. 1e). A greater N availability also significantly increased the DM (Fig. 1f–h) of yield (195%), stubble (205%) and roots (27%). The response of total DM to elevated pCO2 was not affected by the N treatment.
Root–shoot ratio
During regrowth of the field-grown swards the root–shoot ratio of the DM (R : SDM) (Fig. 2a) decreased significantly (P < 0.0001), from 0.94 on day 0 (data not shown) to 0.70 after 14 d of regrowth and to 0.46 by the last sampling after 56 d. The R : SDM was 44% higher at elevated pCO2 than at ambient pCO2 and the response to pCO2 depended on the cutting frequency before the experimental period (pCO2 × frequency, P < 0.05). Infrequently pretreated swards exhibited a 29% (P < 0.01) increase in R : SDM, frequently pretreated swards showed a pCO2-induced increase in the R : SDM of 60% (pCO2, P < 0.0001). The response of R : SDM to elevated pCO2 did not change throughout the regrowth period.
Root–shoot ratio of DM (R : SDM) (a) during 56 days of regrowth of 2-yr-old field-grown Lolium perenne swards at ambient (36 Pa, open symbols) and elevated (60 Pa, solid symbols) pCO2 and after infrequent (four cuts per year, circles) and frequent (eight cuts per year, triangles) cutting treatments before the examined regrowth and (b) during 28 d of regrowth of spaced Lolium perenne plants grown in a controlled environment on sand culture with nutrient solution containing low N (1 mm NO3−, hexagons) or high N (15 mm NO3−, squares) at ambient (35 Pa, open symbols) and elevated (70 Pa, solid symbols) pCO2. Error bars = SE, n = 6.
The R : SDM of regrowing Lolium perenne plants in the growth chamber (Fig. 2b) depended on the sampling date (P < 0.0001). The plants showed the highest R : SDM value (0.48) on the first sampling date of the experiment. Subsequently, they exhibited significantly lower R : SDM values (0.37–0.43). Averaged over the entire regrowth period, elevated pCO2 did not affect the R : SDM (pCO2, ns). Plants grown in a high-N regime exhibited a 52% lower (P < 0.0001) R : SDM than plants grown in a low-N regime. Elevated pCO2 did not affect this relationship (pCO2 × N, ns).
Root–shoot ratio of 14C distribution
The sampling date had a significant (P < 0.01) effect on the root–shoot ratio of the 14C distribution (R : S14C) of the field-grown Lolium perenne swards (Fig. 3); it decreased from 0.12 on the first sampling date after 14 d to 0.08 after 56 d of regrowth. The R : S14C averaged over all the sampling dates was significantly higher (39%) at elevated pCO2 (P < 0.01). The response of R : S14C to elevated pCO2 did not change throughout the experimental period.
Root–shoot ratio of 14C 46 h after pulse labelling (R : S14C) during 56 d of regrowth of 2-yr-old field-grown Lolium perenne swards at ambient (36 Pa, open symbols) and elevated (60 Pa, solid symbols) pCO2 and after infrequent (four cuts per year, circles) and frequent (eight cuts per year, triangles) cutting treatments before the examined regrowth (error bars = SE, n = 6).
The root–shoot ratio of 14C distribution depended on the root–shoot DM ratio, both of which decreased with the increasing length of regrowth (Fig. 4). This relationship did not depend on pCO2, as demonstrated by the comparable slope and intercept of the regression curves of the two pCO2 treatments (t-test for pCO2, ns).
Root–shoot ratio of DM (R : SDM) of Lolium perenne plotted against the root–shoot ratio of 14C 46 h after pulse labelling (R : S14C). Data collected throughout 56 d of regrowth of 2- yr-old field-grown Lolium perenne swards at ambient (36 Pa, open symbols) and elevated (60 Pa, solid symbols) pCO2 (n = 6) (R : S14C = –0.00338 + 0.17 × R : SDM, r2 = 0.82, P < 0.0001). (The R : SDM and R : S14C are derived from 2, 3, respectively.)
Root–shoot allometry
Under field conditions, elevated pCO2 led to a significant displacement of the allometric regression line (Fig. 5a). This was resulted from an increase (P < 0.001) in the intercept b, revealing a greater amount (plus 52%) of root DM per unit shoot DM produced at elevated pCO2 than at ambient pCO2. The slope k of the allometric curves did not significantly change.
Allometry of root and shoot DM obtained (a) during 56 d of regrowth of 2-yr-old field-grown Lolium perenne swards at ambient (36 Pa, open symbols) (Loge(root DM) = 1.95 + 0.08 × Loge(shoot DM), r2 = 0.01, P < 0.785) and elevated pCO2 (60 Pa, solid symbols) (Loge(root DM) = 1.56 + 0.40 × Loge(shoot DM), r2 = 0.11, P < 0.296) and following infrequent (four cuts per year, circles) and frequent (eight cuts per year, triangles) cutting treatments before the examined regrowth (n = 6) and (b) during 28 d of regrowth of spaced Lolium perenne plants grown in a controlled environment on sand culture with nutrient solution containing low N (1 mm NO3−, hexagons) (Loge(root DM) = –0.56 + 1.05 × Loge(shoot DM), r2 = 0.90, P < 0.0001) or high N (15 mm NO3−, squares) (Loge(root DM) = –1.16 + 0.90 × Loge(shoot DM), r2 = 0.96, P < 0.0001) at ambient (35 Pa, open symbols) and elevated (70 Pa, solid symbols) pCO2.
Under controlled conditions (Fig. 5b), there was no difference between the root–shoot allometric relationships of the two pCO2 levels (t-test for both slope k and intercept b: ns), as clearly indicated by the congruent regression lines. The high-N regime led to a change in the root–shoot allometry in favour of the shoot compared with the low-N regime. This difference was manifested by a significantly smaller intercept b of the allometric regression lines (t-test for b, P < 0.01) in the high-N regime.
Discussion
Sustained increase in C allocation to the roots in the field
The two most important results of these experiments are: (1) in well-established 2-yr-old Lolium perenne swards under field conditions there was a sustained increase in C allocation to the roots in response to elevated pCO2; and (2) under controlled conditions no change in C allocation occurred. A higher R : SDM of field-grown grass swards at elevated pCO2 is in line with other experiments (Soussana et al., 1996; Hebeisen et al., 1997; Schapendonk et al., 1997). It was not known, however, whether this is due to (1) an ontogenetic drift, (2) less decomposition of the roots, (3) an extended longevity of roots or (4) a sustained increase in C allocation to the roots. The fact that 14C allocation to the roots remained higher at elevated pCO2 (Fig. 3), even 2 yr after planting, demonstrates that this increase in C allocation to the roots under field conditions was a sustained effect to atmospheric CO2 enrichment.
The intercepts b of the allometric lines (Fig. 5a) of the field experiment data reveal that at elevated pCO2 more root DM per unit shoot DM is produced. This pCO2-related effect did not depend on the change in shoot size, because the slope k of the allometric line was not significantly affected by elevated pCO2. The remarkable absence of an ontogenetic drift in R : SDM confirmed the findings of the 14C labelling. The greater intercept b in the field at elevated pCO2 compared with ambient pCO2 and the constant slope of the allometric line convincingly show that a significant adjustment to elevated pCO2 may occur in swards under field conditions.
The uniformity of the relationship between R : SDM and R : S14C at both pCO2 levels suggests that carbon costs per unit root DM were not affected by elevated pCO2. Combined with the marked increase (by 109%) in root DM at elevated pCO2, our field experiment revealed a strong increase in C costs of the total root system. These increased costs help to explain the discrepancy between a 50% increase in the photosynthesis of single Lolium perenne leaves (Rogers et al., 1998; Isopp et al., 2000) and the small, often statistically nonsignificant, yield response of Lolium perenne swards in the same FACE array (Hebeisen et al., 1997; Suter et al., 2001; see also Fig. 1d).
Higher C costs of the root fraction are in line with findings of other experiments, in which higher carbon costs were indicated by an increase in below-ground respiration related to an increased root DM at elevated pCO2 (Billes et al., 1993; van Ginkel et al., 1997; Schapendonk et al., 1997).
The data on 14C allocation in favour of the root do not enable us to determine whether or not the C is used for (1) new growth of roots in relation to an increased root turnover (Fitter et al., 1996; Loiseau & Soussana, 1999), (2) higher specific root activity (BassiriRad, 2000) or (3) larger amounts of root exudates (Grayston et al., 1998). All these processes may change the quantity (Ineson et al., 1996; Leavitt et al., 1996) and quality (Jongen et al., 1995; Hartwig et al., 2000; Daepp et al., 2001) of the C input into the soil under elevated pCO2. This may be the cause of changes in the number and the population structure of soil and rhizosphere microorganisms (Cotrufo & Gorissen, 1997; Paterson et al., 1997; Sadowsky & Schortemeyer, 1997; Marilley et al., 1999). Such a change could influence the C flux through the plant–soil system, because microorganisms are involved in the decomposition of plant-derived C-containing compounds as well as in the sequestration of C (Berntson & Bazzaz, 1996; Paterson et al., 1997) and nutrient cycling.
Effect of elevated pCO2 on C alloction depended strongly on the experimental conditions
In contrast to the field experiment, elevated pCO2 in the experiment under controlled conditions had no effect on the R : SDM and the allometry. This demonstrates that the response of DM allocation to elevated pCO2 of plants in a sward under field conditions and of individual plants in growth chambers is clearly different.
Which factors are responsible for the observed differences? In the same FACE array the response of Lolium perenne to elevated pCO2 has been shown to be N-limited except when extremely large amounts (112 g m−2 yr−1) of N fertilizer were applied (Daepp et al., 2001). Under a nonlimiting supply of N in the field, the harvest index and the R : SDM of Lolium perenne were not affected by elevated pCO2 (Daepp et al., 2001). Thus, the swards in our field experiment, which were supplied with 56 g N m−2 yr−1, were considered to be N-limited. Elevated pCO2 leads to faster growth of the plant, which increases the need for N. Thus, the increased partitioning of C to the roots may be a reaction of the plant to meet the greater N demand by investing in N-acquiring tissue until the C supply is balanced by acquired N (Brouwer, 1983; Chapin et al., 1987; Schenk et al., 1996; Soussana et al., 1996).
Numerous experiments on N supply and plant growth have been conducted with grass species. Nitrogen shortage led to an increase in C allocation to the root fraction (Lutze & Gifford, 1998; Arredondo & Johnson, 1999; Glimskär & Ericsson, 1999; Mackie-Dawson, 1999; Paterson & Sim, 1999), as observed in our controlled environment experiment, independent of atmospheric pCO2. By contrast, in the high-N treatment in the controlled environment experiment the plants were not limited by N in either of the CO2 treatments. Thus, elevated pCO2 had no effect on the R : SDM. In the low-N treatment in the controlled environment experiment, the growth of the plants was N-limited, even at ambient pCO2. This is clear from the much lower DM production compared with the high-N treatment. However, the increase in the R : SDM caused by low N showed that the roots reacted to environmental changes in the controlled environment. Furthermore, the plants increased their root mass at both levels of CO2 and N throughout the experimental period. Thus, it is unlikely that the pot size was responsible for the lack of an R : SDM response to elevated pCO2 (McConnaughay et al., 1993).
The difference between the unchanged R : SDM in the sand culture in the low-N treatment and the pCO2-induced changes in the field, both at limiting N supply, may be caused by the different supply of nutrients. In the sand culture, the nutrient supply was steady, because of the regular application of nutrient solution. The plants therefore met their additional N demand, which was caused by an increase in the supply of C. By contrast, the field experiment was fertilized only once in 28 d. This may have led to less N being available for several days. Furthermore, a greater amount of nutrients may have been immobilized by soil microorganisms because of a higher C : N ratio in the ecosystem at elevated pCO2 (Díaz et al., 1993).
Another reason for the higher R : SDM at elevated pCO2 under field conditions, but not under controlled conditions, may be a C-sink limitation of the shoot in the dense sward because of, for example, the poor growth of new tillers (Suter et al., 2001). The roots tend to compete less efficiently for C than the shoot (Donaghy & Fulkerson, 1998; Mackie-Dawson, 1999) and, as a consequence, carbohydrate availability for root growth may be better when the shoot sink is weak. This may also explain why the response of R : SDM to elevated pCO2 was stronger in the frequently pretreated swards than in the infrequently pretreated swards. The shoots of single young plants in the controlled environment experiment were not spatially restricted, as demonstrated by the site filling of these Lolium perenne plants, which was at a maximum at both pCO2 levels (Fischer, 1998). Farrar & Jones (2000) hypothesize that C allocation may be controlled by the root and the shoot with certain C- and N-containing compounds as possible signalling factors.
Is the lack of a CO2 effect on the R : SDM, as observed in our controlled environment experiment, a general phenomenon? In many experiments with seedlings an increase in R : SDM was observed (Hunt et al., 1995; Baxter et al., 1997; Lutze & Gifford, 1998). This increase was explained as an ontogenetic effect (Hunt et al., 1995; Lutze & Gifford, 1998; Harmens et al., 2000) related to an increase in the R : SDM with increasing developmental stage (size) combined with a faster growth of the plant under elevated pCO2 (Gedroc et al., 1996; McConnaughay & Coleman, 1999). Thus, if plotted allometrically, this CO2 effect disappears. These plants were very young seedlings which exhibit a strong change in R : SDM with increasing developmental stage. In our controlled environment experiment, however, the plants were already well-established when the measurements were taken (first harvest 52 d after sowing). Thus, there was no important change in the R : SDM during the experimental period, independent of pCO2. As a result, the higher absolute growth rate at elevated pCO2 (Fig. 1e) did not have an effect on R : SDM. There is some evidence that the effect of elevated pCO2 on k is species dependent. For example, Crookshanks et al. (1998) reported a significant enhancement of the k-value of the R : S allometry in Arabidopsis thaliana seedlings. Baxter et al. (1994) found a CO2-induced change in the allometric coefficient in favour of the shoot in Festuca vivipara, while they did not find a change in the R : S allometry of Poa alpina and Agrostis capillaris.
In conclusion, well-established L. perenne swards showed a sustained increase in the C allocation to the roots related to a change in the DM allocation in favour of the roots under elevated pCO2 in the field. In contrast, plants grown in a controlled environment showed no such change in the R : SDM. The causes of the different responses to elevated pCO2 may be N availability, the age of the plants and C-sink strength of the shoot. The sustained greater C allocation to the roots at elevated pCO2 may have pronounced effects on the yield of Lolium perenne as well as on many processes in the soil.
Acknowledgements
We thank W. Wild, W. Wasiak and U. Rossel for technical assistance and H. Blum, R. Bossi, J. Nagy, and G. Hendrey for maintaining the FACE. We specially thank K. Girgenrath and the laboratory of organic chemistry at the ETH for conducting the 14C analyses. M. Schoenberg checked the English. The Swiss FACE was supported by the Swiss Federal Institute of Technology, the Swiss National Energy Research Fund, the Swiss National Science Foundation, the Swiss Department for Energy, the Swiss Department for Agriculture and the Brookhaven National Laboratory (NY, USA).
