Productivity of Trifolium subterraneum and Phalaris aquatica under warmer, high CO2 conditions


Author for correspondence Julianne Lilley Tel: +61 26246 5536 Fax: +61 26246 5399


  •  Despite the importance of grass-legume pastoral ecosystems worldwide, there is little known about the impacts of concurrent increase in temperature and atmospheric CO2 concentration on their productivity.
  •  Pure and mixed swards of subterranean clover (Trifolium subterraneum) and phalaris (Phalaris aquatica) were established under ambient and warmed (+3.4°C) air temperatures, at ambient and 690 µmol mol−1 CO2 concentrations in field tunnels in temperate south-eastern Australia.
  •  Over one year, elevated CO2 increased clover foliage growth in the monoculture by 19%, and by 31% in the mixture. Warming reduced clover monoculture herbage production at ambient CO2 by 28% and reduced the growth enhancement by elevated CO2 to +8%. Forage growth of phalaris monoculture was not affected significantly by either factor. Forage growth of the mixture was increased by 34% in response to higher CO2, but unaffected by warming. Elevated CO2 combined with warming increased forage growth of the mixed sward by 23%.
  •  Concurrent rise in atmospheric CO2 concentration and temperature increased productivity of subterranean clover-phalaris swards. However, longer term effects on species competition and persistence may modify this conclusion.


Global atmospheric CO2 concentration is increasing at a rate of c. 1.8 µmol mol−1 yr−1 (Bengtsson, 1994). A doubling of the preindustrial CO2 concentration is expected by the end of the this century (Watson et al., 1990), with a concomitant increase in global mean temperature of c. 3°C (Houghton et al., 1990). CO2 enrichment of the atmosphere has been reported to cause various responses among plants species, but the majority of experimental evidence indicates that plant growth is stimulated by CO2 enrichment in the field and in controlled environments.

The effect of CO2 enrichment on plant growth may be strongly affected by factors such as air temperature, plant growth stage and availability of resources such as water, radiation and nutrients (Drake et al., 1997). Plant development is generally accelerated by increased temperature, while growth rate increases up to an optimum temperature range and declines at higher temperatures (Rawson, 1992). Several authors have reviewed the effects of elevated CO2 and air temperature on plants (Idso et al., 1987; Eamus, 1991; Long, 1991; Rawson, 1992). Newton et al. (1994) suggested that the effect of CO2 in temperate regions is seasonal because of the interaction between CO2 and temperature. Aboveground response of pasture species to CO2 has been found in some studies to occur only at higher temperatures (> 19°C) (Newton et al., 1994), while in other species a positive response to CO2 was weakened at higher temperature (Ferris et al., 1996; Nijs & Impens, 1997).

In many studies of pastures containing C3 grasses and legumes, growth in elevated CO2 altered species composition in favour of the legume (Newton et al., 1994; Clark et al., 1997; Hebeisen et al., 1997a; Schenk et al., 1997a). More generally, elevated CO2 has been observed to stimulate the growth of dicots relative to monocots in temperate pasture communities (Potvin & Vasseur, 1997).

There are many studies from controlled environments examining temperature and CO2 interactions. However there is a need to complement these experiments with field studies, where realistic diurnal and seasonal temperature and radiation fluctuations occur. Despite numerous studies of perennial ryegrass (Lolium perenne) and white clover (Trifolium repens), there are no studies on the effects of elevated CO2 and air temperature on the temperate pasture ecosystems of south-eastern Australia. Consequently we studied the effects of elevated CO2 and increased temperature on phalaris (Phalaris aquatica cv. Holdfast), a C3 grass, and the self-seeding, summer-dormant, annual subterranean clover (Trifoliumsubterraneum cv. Mt Barker) grown as monocultures and as a mixture. Phalaris is the most common sown grass on the tablelands of New South Wales, where it is generally grown together with subterranean clover in pastures for grazing sheep and beef cattle. The area occupied by phalaris-based pastures across southern Australia is c. 1.5 million hectares (Carlson et al., 1996). This paper reports on productivity and competitive relationships of pure and mixed subterranean clover/phalaris swards grown in the field in Canberra, Australia, under two temperature and two CO2 regimes. Aspects of pasture quality and nitrogen fixation of these swards are reported by Lilley et al. (2001).

Materials and Methods

Description of the temperature gradient tunnels

The six tunnels were modified from square-section tunnels, 1.25 m wide and 1.25 m high, described by Gifford & Rawson (1993) and Rawson et al. (1995). Changes consisted of structural modifications and changing the plastic on the top surface to clear Teflon (S. W. Plastics Ltd, Crayford, Kent, UK) for improved strength and superior durability, light and UV-B transmitting properties. The Teflon film transmitted 96% of visible solar radiation and lost no clarity during the experiment. Structural modifications, including placing the fan-board centrally instead of at the inlet, enabled the tunnels to be operated with a step change in temperature, rather than with a continuous gradient. This modification eliminated the problem of occasional loss of temperature gradient between ends of the tunnel under certain environmental conditions. As described by Rawson et al. (1995), there was no discernible CO2 gradient. The inlet section and controlling and heating zones were covered with UV stable polyethylene sheet (Greenland Film, Flora-Guarde Pty Ltd, NSW, Australia). Experimental sections had polycarbonate sheet sides (Rawson et al., 1995). Light transmission properties of the polycarbonate sheet are given in Gifford & Rawson (1993).

Total tunnel length was 11.8 m. The layout is illustrated in Fig. 1. Along the length of each tunnel there was: a short (0.8 m) air inlet section where, in the case of the elevated CO2 tunnels, pure CO2 was injected and mixed; a ‘field temperature’ experimental section (4.9 m), where air temperature was close to ambient; a controlling zone (3 m) which contained the electronic controls and data logging equipment, fans, and heaters in a heating compartment; and a final ‘warmed’ experimental section (3.1 m) in which the air outlet was located.

Figure 1.

Layout of tunnels used in the study viewed from above. Sections or equipment are marked as follows: (1) Inlet chamber; (2) ‘field temperature’ section; (3) controlling and solar heat absorbing section; (4) ‘warmed’ section; (a) inlet vent; (b) equipment cupboard; (c) partition containing fans; (d) heating coils; and (e) outlet vent. X denotes positions of thermocouples measuring air temperature; + denotes position of thermocouples in the soil.

Air was taken into the inlet section through an upward-facing gap (0.15 × 1.25 m) and passed through two vertical baffles, with perforations at the bottom of the first baffle, and at the top of the second baffle, to maximize mixing of the CO2. The perforated areas extended across the width of the chamber, were 0.3 m high and consisted of holes 3-cm in diameter such that 70% of that area was holes. In elevated CO2 tunnels, CO2 was introduced between these two baffles. The CO2 was injected in pulses controlled by a solenoid valve into a pressurized plastic drum that slowly released the gas through a capillary to dampen fluctuations in CO2 concentration. The CO2 from the capillary was injected into the back of a small mixing fan and mixed in the inlet chamber with ambient air before it passed through the second perforated baffle and into the ‘field temperature’ section. CO2 concentration was sensed by CO2 space-monitors (ADC-2000, Analytical Development Co. Ltd, Hoddesdon, UK) in the middle section of each elevated CO2 tunnel, and one ambient CO2 tunnel. Monitor outputs were input to electronic controllers that determined the pulse rates of the solenoids in the inlet section in proportion to the difference between target CO2 concentrations and those measured in the controlling section.

The soil surface of the controlling zone was covered with black plastic to prevent soil-moisture evaporation, and to act as a solar heat absorber, thereby increasing radiative heating of the downstream section. The controlling electronic circuitry, data loggers and CO2 analysers were installed in this section, in a shaded cupboard. Three fans were installed in a vertical board, separating the upstream from the downstream section and positioned two thirds of the way through the central controlling section. The fans drew air from the ‘field temperature’ section and blew air into the downstream heating compartment. The warmed air was then passed through a baffle, perforated about 70% with numerous 3-cm diameter holes to distribute the heated air across the tunnel cross-section. Airflow was varied continuously by electronically changing the fan speed between 30% and maximum speed to achieve the 4°C target temperature difference. Heating was added as necessary in conditions of low radiation. The heating section contained a heating element, placed at the same level as, and downstream of the fans, so that air was warmed under continuous electronic proportional control as it passed over the heating coils. The vertical air exit vent was protected by a raised stainless steel cone, which prevented rain and wind gusts from entering the tunnel.

Each tunnel was irrigated with a microjet spray irrigation system, positioned 30 cm above the soil surface. Each section contained nine individually adjustable sprays to ensure uniform water application. Irrigation was applied regularly so that water was not limiting to growth at any stage of the experiment.

Temperatures were monitored using well-shielded, aspirated thermocouples. Temperature was measured twice per minute and hourly averaged readings were recorded using a data logger (DataTaker 600, DataTaker Pty Ltd, Rowville, Victoria, Australia). Two thermocouples were located centrally in each experimental section of the tunnel, 50 cm above the soil surface, so that they were close to the centre of the tunnel and above the pasture canopy at all times (Fig. 1). On average, temperature within the ‘field temperature’ section varied < 0.2°C between the two thermocouples within the section. In the warmed section the average temperature difference between thermocouples (0.8 m apart) was 0.1°C. Soil temperatures were measured using a thermocouple buried 5 cm below the surface in the middle of the ‘field temperature’ and ‘warmed’ experimental sections. Temperature control used the difference between two continuously sampled shielded and ventilated thermocouples positioned in the opening of the inlet chamber and 1 m into the warmed section (Fig. 1).

Description of the experiment

The six tunnels were established at the Ginninderra Experiment Station (GES) (149°06′ E, 35°12′ S; elevation 600 m), near Canberra, Australian Capital Territory, Australia. The site had been a grazed pasture for over 20 yr. To control weeds and improve site uniformity, the site was fertilized and sown to green manure crops of millet in November 1993 and oats in July 1994, which were incorporated in March and November 1994, respectively. The soil was cultivated in August and September 1995 before erection of the tunnels. The soil was a yellow podzolic (Gn3.85, Northcote, 1971). The surface 10 cm contained 1.61% total C, 0.17% total N and had a surface pH of 4.3 in CaCl2. Total C and N decreased down the soil profile while pH increased. Available (Bray & Kurtz, 1945) phosphorus was 18.3 kg ha−1 in the surface 10 cm. Mineral nitrogen in the surface 30 cm was 111 kg ha−1 before sowing. A basal fertiliser application of superphosphate and molybdenum (28 S, 23 P, and 0.075 Mo kg ha−1) was applied at sowing.

Two temperature and two CO2 regimes were imposed in a split-plot design with CO2 as main plots (whole tunnels), temperature as subplot (within tunnels), and with three replications. Each temperature × CO2 subplot (2.45 m × 1.25 m) contained three sward types. These were phalaris monoculture, clover monoculture, and a 50 : 50 mixture of these species on an area basis using a substitution design. Nonenclosed comparison plots, outside the tunnels, were also established to examine the effect of the plastic tunnel on the pasture at ambient CO2 and ambient temperature. The two temperature regimes followed diurnal and seasonal temperature fluctuations. These were ‘field temperature’ which was close to the ambient temperature and a ‘warmed’ treatment which was controlled to a target 4°C above ‘field temperature’. CO2 treatments were ‘ambient’ and ‘elevated CO2’ which was targeted to 700 µmol mol−1.

Pure and mixed swards of phalaris and subterranean clover were established in the ‘field temperature’ and ‘warmed’ section of each of the tunnels. Seeds were sown into a fine seed bed on 14 December 1995 (early summer), in rows 8.5 cm apart, to create 14 rows running the length of the tunnel, and two rows on the outside of the tunnel on each side. Seed density was selected to obtain an equal plant density of clover and phalaris. The total population density target was 235 plants m−2. The mixture was sown in a checkerboard pattern made up of alternating rows and a row-segment length of 23 cm for each species. Swards were well watered and hand weeded for the duration of the experiment.

Environmental conditions

During the experiment, average CO2 concentration was 380 µmol mol−1 in the ambient tunnels and 690 µmol mol−1 in the elevated CO2 tunnels (Fig. 2). Elevated CO2 concentration was within the 600–800 µmol mol−1 range for more than 90% of the time. The main departures from target occurred when tunnel side-panels were removed to work on the plants. Total solar radiation was high (15.4 MJ m−2 d−1) early in the experiment during summer (Dec–Feb), and declined through autumn and winter (Jun–Aug) to 6.0 MJ m−2 d−1. Average daily air temperatures in the ‘field temperature’ and ‘warmed’ treatments followed a similar pattern. The average difference between the two sections was 3.4 ± 0.3°C. Diurnal variation in temperature ranged from 1 to 26°C in all sections and outside the tunnels, with an average variation of 14°C (data not shown). Temperatures in the ‘field temperature’ treatment were 0.3–1.4°C above that outside the tunnels. Average soil temperatures followed a similar pattern to air temperatures. The average soil temperature in the ‘warmed’ treatment was only 1.9°C greater than that in the ‘field temperature’ section while outside the tunnels soil temperature averaged 0.8°C less.

Figure 2.

Average environmental conditions for the intervals preceding each harvest during the study. (a) Average CO2 concentration for ambient (squares) and elevated (triangles) treatments; (b) total incident solar radiation outside the tunnel; (c) air temperature 0.5 m above the soil surface; and (d) soil temperature for nonenclosed comparison plots (open circles) field temperature (open diamond) and warmed (closed diamond) tunnel sections.



The sampled area for each plot was 10 rows of 40 cm length (0.34 m2). Guard zones consisted of two rows within the tunnel plus two rows outside the tunnel on each side, and at least 15 cm of row length, depending on position of the plot within the section. Swards were harvested on Jan 12, Jan 30, Feb 15, Mar 6, Apr 2, May 15, July 25, Sept 4, Oct 3, Oct 25, and Nov 26 1996 (29, 47, 63, 83, 110, 153, 224, 265, 294, 316, and 348 d after sowing). Swards were cut with hand-held electric grass clippers 7 cm above ground level, separated into plant species, and d. wt determined. Frequency and height of cutting were chosen to represent grazing of the pasture, yet ensure adequate regrowth during each harvest interval. At harvest 10 (Oct 25), half of the harvested area was also cut to ground level to establish the biomass between the soil surface and the cutting height of 7 cm (plant bases). At the final harvest (no. 11, Nov 26) the harvested area was 5 rows by 40 cm (0.17 m2) and biomass above the 7 cm cutting height (herbage) and plant bases were sampled separately. Plant bases were also separated into plant species and dead and green material. Following the final harvest a soil sample of 0.19 m × 0.255 m × 0.15 m deep was dug from each plot in the area where the live biomass had just been removed. Roots were washed free of soil (species were not separated) using a 2-mm sieve and d. wt determined after oven drying.


Photosynthetic rate of attached leaves was measured on 10–12 November (between harvests 10 and 11) for a range of external CO2 concentrations with a Li-Cor 6400 Portable Photosynthesis System (Li-Cor Inc., Lincoln, NB, USA). Measurements were made at a leaf temperature of 25°C, and nonlimiting irradiance of 2000 µmol quanta m−2 s−1 provided by a red/blue LED (light emitting diode) light source. Phalaris in the monoculture and the mixture and clover plants grown in monoculture were sampled. One leaf was measured in each replicate of each temperature × CO2 treatment combination. The leaves measured were fully expanded and near the top of the canopy in a fully exposed position. Assimilation rate was plotted against intercellular partial pressure of CO2 and coefficients of response curves fitted to the biochemically based model of von Caemmerer & Farquhar (1981) were calculated for each leaf. Assimilation rate at external CO2 concentrations of 380 and 690 µmol mol−1 was then calculated from each curve.

Statistical analysis

Data were analysed by ANOVA using a split plot design (CO2 as whole plot and temperature as subplot treatments) and incorporating harvest as a repeated measure where appropriate (Steel & Torrie, 1980), using PROC GLM of the SAS statistical software package (SAS, 1989). Sward types were analysed separately. F-tests were used to detect significant differences between treatments. Ratios and percentage changes presented in tables were tested using pairwise comparisons between the means of the raw data of relevant treatments.


Cut herbage production

Seasonal effects

Growth rate of all treatments followed seasonal changes in temperature and radiation. They were highest in summer, soon after establishment, when temperatures and radiation were high (Fig. 3). For all harvest intervals, growth rate of herbage (above 7 cm cutting height) was greater for clover than for phalaris, and the growth rate of the mixture was similar to that of clover monoculture. The differences between treatments in accumulated herbage over the duration of the experiment are shown in Fig. 4.

Figure 3.

Forage growth rates above the 7 cm cutting height (g m−2 d−1) (a) clover monoculture; (b) phalaris monoculture; and (c) mixed sward of clover and phalaris, grown under a combination of two CO2 treatments (squares – ambient; trianlges – elevated) and two temperature treatments (open symbols – field temperature, closed symbols – warmed temperature). F-tests indicate the following main effects and interactions were significant (P < 0.05): (a) CO2, temperature, harvest; (b) harvest; (c) CO2, harvest, CO2 × harvest.

Figure 4.

Total shoot biomass (above the 7 cm cutting height) removed between sowing and final harvest for (a) clover monoculture (b) phalaris monoculture, and (c) mixed sward of clover and phalaris, separated into clover (grey columns) and phalaris (black columns) components. Temperature treatments: F, field temperature; W, warmed air. F-tests indicate the following main effects and interactions were significant (P < 0.05): (a) CO2, temperature; (b) nil; (c) total – CO2; clover component – CO2; phalaris component – CO2 × temperature.

Phalaris and subclover are typically sown during autumn, and the main growing season is autumn through to spring. Sub-clover sets seed and senesces in early summer and phalaris, a perennial, becomes dormant during hot, dry summer months. For reasons that are unclear, plants in the nonenclosed comparison plots were slow to establish out of season relative to plants in the tunnels. However, from autumn onwards, when all plants were well established, growth rates and other characteristics of clover and phalaris in both monoculture and mixture in the nonenclosed comparison plots were similar to those in the ambient CO2, ‘field temperature’ treatment. Therefore, there does not appear to be an effect of the tunnel on plant growth. To simplify description of the results, nonenclosed comparison data are not presented.


Warming decreased the growth rate of clover herbage under ambient CO2 at all harvests, even during winter months (Fig. 3) when ‘field temperature’ averaged 7°C (Fig. 2). The overall warming effect on clover herbage growth was −28% (Table 1). For phalaris, warming tended to increase growth rate, but not significantly (Table 1). Warming did not alter the overall productivity of the mixture (−3%), although the composition of the mixture was radically altered. In the mixture, clover production was substantially reduced (55%) by the warming, while phalaris grew extremely competitively, producing more than double the cut herbage of the field temperature, ambient CO2 treatment.

Table 1.  Percentage change in cumulative herbage (above a cutting height of 7 cm) due to elevated CO2 and warming for swards of clover and phalaris in monoculture and mixture over the duration of the experiment
Effect ofCloverPhalarisTotalCloverPhalaris
  • *

    , significant deviation from zero at P < 0.05.

CO2 elevation at ‘field temperature’  19* −534*  31*  40*
CO2 elevation in ‘warm’ treatments  49*  227159* −35
Warming at ambient CO2 −28*  9 −3 −55*110*
Combination of elevated CO2 and warming relative to ‘field temperature’ and ambient CO2    81123*  17*  36*

Effect of elevated CO2 alone and with concurrent warming

Cumulative clover herbage production at field temperatures was increased by 19% by the elevated CO2, but by only 8% when increased temperature was combined with elevated CO2 concentration (Table 1, Fig. 4). The cumulative herbage production under the warmed treatment was increased 49% by elevation of the CO2. The greatest response of clover to CO2 was observed early in the experiment, and was highest at the May harvest, when growth rate under elevated CO2 was more than doubled in both temperature treatments, and declined during winter and spring (Fig. 3). There was no significant CO2 effect on growth of phalaris at either field or warmed temperatures (Table 1).

In the mixture, response of accumulated herbage to CO2 was +34% at field temperature, but was a nonsignificant +27% in the warmed treatment. When both temperature and CO2 were elevated, clover and phalaris components each increased relative to ambient CO2-field temperature, although the response of phalaris was greater. The overall response of the mixture to the combined warming and CO2 elevation was a 23% increase in accumulated herbage.

Plant bases

Plant bases below 7 cm (excluding any leaves and stems that had died and decayed) accumulated during the experiment and were sampled at harvests 10 and 11. For clover bases there was a positive response to CO2 at both harvest 10 and 11, while the negative response to temperature at harvest 11 was not significant (Fig. 5). The net accumulated plant bases of phalaris showed a positive response to elevated CO2 and negative response to warming at harvest 10, but no significant differences were apparent at harvest 11. In the mixture there was a response of accumulated plant bases to CO2 at harvest 11, and a negative response to warming of the clover component despite no overall response to warming. Biomass of plant bases increased for both clover and phalaris in the mixture and monocultures between harvest 10 and harvest 11. For phalaris this consisted mainly of enlarging tiller bases, while for clover this contained a proportion of setting seed. The increase in weight of clover bases between harvest 10 and 11 was less at warm temperature and was related to the beginning of senescence, which was advanced by warm temperatures, and delayed by elevated CO2.

Figure 5.

Biomass (g m−2) of plant bases (0–7 cm) accumulated (a), (b), (c) to harvest 10 (25 October); and (d), (e), (f) to harvest 11 (26 November) for clover monoculture, phalaris monoculture, and mixed sward of clover and phalaris, respectively (clover (grey columns) and phalaris (black columns)). Temperature treatments: F, field temperature; W, warmed air. F-tests indicate the following main effects and interactions were significant (P < 0.05): (a) CO2; (b) CO2, temperature; (c) nil; (d) CO2; (e) nil; (f) total – CO2; clover component – CO2, temperature; phalaris component – nil.


The total amount of recoverable root biomass in the top 15 cm of soil for phalaris was 2.5 times higher than that of clover in all treatments, while the total root biomass of the mixed sward was intermediate (Fig. 6). The final root mass of pure clover, pure phalaris and the mixture were all increased by elevated CO2 (about 35%) and consistently reduced by warming (40%), although the temperature effect was only significant for the phalaris monoculture. In the combined elevated CO2 plus warming treatment, root biomass was similar to ambient CO2-field temperature treatment, again demonstrating that the warmer conditions negated the advantage of CO2 increase. In each sward type, root biomass in the nonenclosed comparison was not significantly different from that in the ‘field temperature’ treatment (data not presented).

Figure 6.

Root biomass (g m−2) to a depth of 15 cm, harvested on 26 November for (a) clover monoculture; (b) phalaris monoculture; and (c) mixed sward of clover and phalaris. Temperature treatments: F, field temperature; W, warmed air. F-tests indicate the following main effects and interactions were significant (P < 0.05): (a) CO2; (b) CO2, temperature; (c) CO2.

At harvest 11 root biomass contributed around 13% of the biomass in clover, 40% of the biomass in phalaris, while in the mixture it was intermediate (26%) (Table 2). Only in phalaris were there significant treatment effects on root : total biomass ratio. This was caused by the responsiveness to CO2 and temperature of the roots, and the lack of response in the above ground biomass. For clover and the mixture there were no significant effects of environment, although there was a very small increase in proportion of roots under elevated CO2.

Table 2.  Recovered root : total biomass ratio at harvest 11, for clover and phalaris monoculture swards and a mixed sward under four combinations of CO2 concentration and temperature regime. Main effects of CO2 and temperature were significant for phalaris (P < 0.05)
 Ambient CO2Elevated CO2
 Field temperatureWarmedFieldWarmed temperature

Whole-plant responses

Accumulated herbage, plus plant bases and roots recovered at harvest 11 is shown in Fig. 7. The CO2 and temperature response ratios for this cumulative biomass are summarized in Table 3. These ratios were relatively consistent at each harvest throughout the experiment (data not shown). At field temperatures, the response of pure clover to elevated CO2 was +19% for all fractions of the plant while for the mixture it was around +30%. There was a negative response to warming in clover (−26%) and the mixture (−15%), and no significant response to the combined CO2 and warming treatment (+3% in clover and +10% in the mixture). For phalaris, the total cumulative biomass responses were small (−6 to +9%) and not significant, since aboveground biomass was unresponsive and the root fraction made up only 14–29% of the cumulative biomass (Fig. 7).

Figure 7.

Total biomass (g m−2) of (a) clover monoculture; (b) phalaris monoculture; and (c) mixed sward of clover and phalaris. Components consist of roots (black columns), plant bases (0–7 cm) at harvest 11 (white columns), and cumulative shoot biomass above 7 cm (grey columns). Temperature treatments: F, field temperature; W, warmed air. F-tests on total biomass indicate the following main effects and interactions were significant (P < 0.05): (a) CO2, temperature; (b) nil; (c) CO2.

Table 3.  Effects of CO2 increase and/or warming on cumulative shoot plus root biomass of swards of clover and phalaris grown as monocultures and as a mixture over a 12-month period. The values shown are the ratios of biomass under elevated CO2, warmer temperature or both relative to biomass in the field temperature, ambient CO2 treatment
  +  CO2 response ratio at field temperature + Temperature response ratio at ambient CO2 +  CO2 & + temperature response ratio
 Biomass componentBiomass componentBiomass component
 > 7 cm> 0 cmtotal> 7 cm> 0 cmtotal> 7 cm> 0 cmtotal
  • *

    , significant deviation from unity at P < 0.05.


Overall, warming reduced cumulative biomass of all components of the clover plant by about 25% (Table 3) and by 6% in total phalaris biomass. For phalaris the cumulative response of cut herbage (above 7 cm) to warming was a 9% increase, but that was nonsignificant and there was a large negative response in the root fraction. The effect of warming on mixture growth was negative, with a slight reduction (3%) in cut herbage, but a large reduction in both the 0–7 cm layer (36%) and the root biomass (37%) (Fig. 7), leading to a reduction in total biomass of 15%. The combination of CO2 and warming treatments resulted in only a small response in clover, and a slightly larger response in phalaris and in the mixture.


Net photosynthesis rates of leaves (Table 4) were calculated from photosynthetic curves for the intercellular concentrations experienced at external CO2 concentrations of 380 and 690 µmol mol−1. There was no significant difference between temperature treatments, so means of these treatments are presented in the table. Carbon dioxide assimilation rates were higher for clover than for phalaris. The calculated CO2 assimilation rate for clover plants grown and measured at elevated CO2 was 12% higher than for clover grown and measured at ambient CO2, while the increase was 20% for phalaris monoculture and 50% for phalaris grown in the mixture. There were no significant differences in leaf area between treatments at harvests 9 or 10. Species differences were similar to those observed for biomass, with a mean leaf area index for pure clover, pure phalaris and the mixture of 3.4, 1.9 and 3.8 at harvest 9 and 4.7, 2.7 and 4.8 at harvest 10, respectively.

Table 4.  Net CO2 assimilation rates (and standard errors of mean) for clover and phalaris leaves calculated at atmospheric CO2 concentrations (Ca) of 380 and 690 µmol mol−1, from CO2 concentration response curves measured at nonlimiting irradiance and 25°C. Measurements were taken between harvests 10 and 11
  Assimilation rate (µmol m−2 s−1) calculated for measurement Ca of:
SpeciesGrowth CO2 concentration380 µmol mol−1 CO2690 µmol mol−1 CO2
CloverAmbient22.6 ± 0.929.4 ± 1.1
monocultureElevated19.0 ± 0.825.3 ± 0.9
PhalarisAmbient15.8 ± 1.324.8 ± 1.9
monocultureElevated12.0 ± 0.719.0 ± 1.0
PhalarisAmbient15.8 ± 1.121.3 ± 1.1
mixtureElevated16.9 ± 1.523.7 ± 2.0


Phalaris established more quickly after sowing, although by 8 wk (15 Feb) clover began to dominate the mixture in all but the ambient CO2, warm treatment (Fig. 8). In the warmed, ambient CO2 treatment, phalaris dominated throughout the experiment, but the proportion of clover increased between the second and sixth harvests, reaching 48%. After this, the proportion of clover decreased during the winter months, increased again during spring, but declined rapidly when it began to senesce in summer. In the 0–7 cm layer of the sward the dominant component was clover, where it contributed 65–85% of the sward in all treatments at harvest 10 (Fig. 8).

Figure 8.

The proportion of clover biomass in a mixed sward of clover and phalaris grown for 12 months under a combination of two CO2 treatments (squares – ambient; triangles – elevated) and two temperature treatments (open symbols – field temperature, closed symbols – warmed temperature). Values for plant bases at harvest 10 and 11 are shown at right. F-tests indicate a significant CO2–temperature interaction (P < 0.05).

The mixture was 11–43% more productive than the average of the phalaris and clover monocultures, and had a similar productivity to the more productive monoculture for all components of biomass (Table 5). In this experiment, the more productive monoculture was always clover. Only in the case of the warmed, ambient CO2 treatment was the mixture more productive than clover (by 24%) in the above 7 cm fraction. On a total above-ground basis, the mixture was slightly less productive than the more productive monoculture, although this was only significant in the field temperature, ambient CO2 treatment. On a total plant basis there was no difference between productivity of the mixture and the more productive monoculture.

Table 5.  Effects of CO2 and warming on yield of mixed swards relative to those of clover and phalaris monocultures. Values are for the cumulative pasture yield over 11 harvests
  Ambient CO2Elevated CO2
Ratio of yield of mixture to:Pasture componentField temperatureWarmedField temperatureWarmed
  • *

    , significant deviation from unity at P < 0.05.

Average yield of> 7 cm1.18*1.34*1.43*1.33*
monocultures> 0 cm1.111.14*1.24*1.18
 total plant1.091.13*1.21*1.15
Yield of highest> 7 cm0.931.24*1.051.06
yielding> 0 cm0.86*0.990.910.93
monoculturetotal plant0.911.030.960.96


Clover monoculture

Total biomass production of the pure clover sward was 20% greater under elevated CO2 than ambient CO2 at field temperatures and was reduced 26% by warming. Other field studies have found similar responses of field-grown white clover monocultures to a doubling of CO2 concentration over 1–2 yr (Hebeisen et al., 1997a; Schenk et al., 1997b). However Hebeisen et al. (1997a) found that the response declined in the third year, suggesting longer-term studies may result in new equilibria. Warming negated the response to elevated CO2 in this experiment. Other studies also have observed small responses to warming at doubled CO2 concentration, but not a reduction in biomass (Ryle & Powell, 1992; Manderscheid et al., 1997). This lack of a positive growth response to warming in winter may relate to the balance between whole-plant respiration and photosynthesis at low winter radiation. At photosynthesis-limiting irradiance, the respiratory load on closed swards of subterranean clover increases with temperature more than net photosynthesis, even in cool conditions (Fukai & Silsbury, 1977).

The primary mechanism of CO2 effects on growth of well-watered vegetation is expected to be via net CO2 exchange modified by positive and negative feedbacks such as changes in leaf area index and metabolic down-regulation of CO2 fixation. Net leaf CO2 exchange measurements indicate that potential assimilation rate (nonlimiting light and temperature) was 12% higher at elevated CO2. This is somewhat less than the 19% overall cumulative increase of total biomass as a result of CO2 enrichment (Table 3). Given that most leaves were not operating at light saturation, these relative enhancements of leaf net photosynthesis and of growth are reasonably comparable. Such comparability of relative response in leaf CO2 exchange and growth are to be expected given that the carbon loss through plant respiration relative to photosynthesis tends not to be influenced much by CO2 or temperature (Gifford, 1995) and that with the frequent defoliation, loss of leaf through death and decomposition would be minor. A study by Newton et al. (1996) showed similar results, with 33% more carbon fixed at doubled CO2 concentration. Manderscheid et al. (1997) also showed that radiation-use efficiency of white clover increased 44% with approximate CO2 doubling.

The subterranean clover began to senesce in early summer with herbage growth rate declining by the November harvest. Senescence of foliage was advanced by warming, and delayed by elevated CO2. However, biomass of clover bases increased significantly between harvests 10 and 11 in all treatments, when the plant was setting seed (Fig. 5). The greater biomass of clover bases under elevated CO2 was likely to have increased the number of shoot apices available for infloresence production and therefore more seed may have been produced in this treatment if the experiment had continued (Collins, 1978). Greater seed production would impact on plant populations in subsequent seasons. Newton et al. (1994) suggested that natural fluctuations in species content of pasture swards, due to changes in senescence patterns, could be altered by CO2 elevation and warming. For self-regenerating, annual pasture species such as subterranean clover, environmental conditions at the time of germination and establishment can have large effects on their relative abundance in the pasture (Rossiter, 1966), and these may well be affected by climate change.

Phalaris monoculture

In contrast to clover, the phalaris was unresponsive to the environmental treatments. There was a small positive response to CO2 in growth rate of phalaris herbage early in the experiment, but during the winter this trend was reversed and spring growth tended to be lower under elevated CO2 (Fig. 3). Meier & Fuhrer (1997) observed that the positive effect of CO2 on growth of a legume-grass mixture declined with time when N supply was limited. Throughout the present experiment the mineral N content of the soil was high (111–83 kg N ha−1). Nevertheless the N concentration of the cut herbage (mean 2.4 and 2.0% at ambient and elevated CO2, respectively; Lilley et al., 2001) was only moderate. As frequently cut phalaris herbage can sometimes attain N concentrations of over 3% (Clements, 1973) it is unlikely that these swards were N saturated. Thus it is possible that the growth responses to CO2 and temperature might have been larger had their N status been higher. Nevertheless, reported responses of perennial ryegrass monocultures to elevated CO2 of up to 9% over a season are also lower than for clover species (16–42%; Hebeisen et al., 1997a,b; Schenk et al., 1997b), but responses are higher (20%) for grass mixtures (Mortensen, 1997). It remains unclear whether grasses are inherently less responsive than nodulated clovers to CO2 and temperature or whether they would be similar if N saturated.

Photosynthesis was measured to determine whether the lack of response in forage growth of phalaris had a photosynthetic basis. These measurements indicate that assimilation rate of phalaris at optimal conditions (nonlimiting light and temperature) was 20% higher when grown and measured at elevated CO2 than when grown and measured at ambient CO2 in the monoculture and 50% higher in the mixture (Table 2). Therefore lack of response to CO2 enrichment of phalaris growth (above the 7 cm defoliation level) cannot be attributed to a lower photosynthetic rate. A significant increase in plant bases (at harvest 10 only) and roots under elevated CO2 indicate some extra carbon was fixed and allocated to these fractions of the plant. Overall plant response (including herbage, plant bases and roots) to elevated CO2 was 9% (Table 3), not as large as that of clover and significantly less than light-saturated assimilation rate. The reason for this low, whole-plant growth response relative to the photosynthetic response is not obvious to us. In part it might be attributable to a greater partitioning of biomass below ground (including to root turnover, which was not measured) in the phalaris. Biomass partitioned to root instead of leaf does not contribute to a positive feedback through extra leaf expansion. Large biomass of roots and plant bases will have implications for regrowth of this perennial plant in the following autumn season. Soussana et al. (1996) observed a 45–52% increase in root biomass of a pure ryegrass sward in elevated CO2, at the expense of above ground biomass while Hebeisen et al. (1997a) suggested that increased allocation to perennial ryegrass roots was related to increased N limitation under elevated CO2. Mild N limitation may have contributed to a lower photosynthetic rate in the monoculture than in the mixture. We hypothesize that a marginal N limitation might have been involved in the low growth response to elevated CO2 by causing mobilization of leaf N from ageing or partially shaded leaves sooner in the elevated CO2 plants than in the ambient CO2 plants, to support new leaf and root growth. However Meier & Fuhrer (1997) observed that the negative effects of N limitation were reduced by elevated CO2.

Experimental warming increased cumulative phalaris herbage over the year by 9% (Table 1), but there was a large negative response to warming in the root fraction (−40%). Since roots made up 14–29% of cumulative biomass of phalaris (Fig. 7), there was a negative effect of warming (−6%) on total phalaris biomass (Table 3). Relatively few field experiments have been conducted under the range of air temperatures found in our study (daily average 7.1–20.5°C). In our study, warming did not increase growth rate to the level obtained at similar temperatures in the field temperature treatment earlier in the season. As discussed above, it appears that low radiation imposed a limitation on growth rate of phalaris during the winter. Growth rate outside the tunnel, where incident radiation was greater, was 10% lower than in the field temperature treatment in the tunnel during the same period, but the comparison is confounded by slightly lower average temperatures and the occurrence of radiation frosts outside the tunnel.

Only for the phalaris were there significant treatment effects on root : total biomass ratio (Table 2). This reflected the responsiveness to CO2 and temperature of phalaris roots, the large proportion of biomass in roots and the lack of response in the above ground biomass. The large response of phalaris root biomass to elevated CO2 (Fig. 6) is consistent with the 50% root increase observed in ryegrass/pasture turves by Ross et al. (1995). They also observed that a combination of elevated CO2 and 6°C warming decreased root mass by 6% which is similar to the small response (−6 to +4%) in our study (Fig. 6).

Mixtures and competition

In general, growth rate of the mixture was similar to that of clover monoculture. A notable exception was the mixture in the ambient CO2 warmed treatment, which yielded 24% more herbage than the clover monoculture (Table 5). It is considered that for a mixture to produce greater biomass than that of the most productive monoculture (of species comprising the mixture) it must exploit the environment more completely than the monoculture (Trenbath, 1974). This phenomenon, termed over-yield, has intrigued ecologists and agriculturalists looking for relationships between species richness and productivity. However, there are relatively few examples of over-yield occurring, even in the case of grass-legume mixtures (Trenbath, 1974; Garnier et al., 1997). Rather, the productivity of the mixture is generally similar to the productivity of the most productive monoculture, as is true for most of the data presented here. In the ambient CO2, warmed treatment the over-yield of the mixture seems to be a result of the reduced growth of the clover and the enhanced growth of the phalaris, together with a N sparing effect (sensuHerridge et al., 1995) conferred to the phalaris by the ability of clover to fix much of its own N (Lilley et al., 2001).

Clover dominated the mixture for all treatments from 47 d after sowing, except the warmed treatment at ambient CO2 (Fig. 8). Elevated CO2 did not significantly alter the proportion of clover in the mixture. This contrasts with results of several others who found that elevated CO2 significantly affected competition within the pasture, generally increasing the level of legume dominance (Hardacre et al., 1986; Newton et al., 1994; Clark et al., 1995; Schenk et al., 1997a). In our study, the increase in cumulative herbage in the mixture was 34%, made up of a 31% increase in subterranean clover and 40% increase in phalaris (Table 1). Several studies have shown that legume dominance was enhanced by elevated CO2 in conditions of limited N supply (Schenk et al., 1997a; Soussana et al., 1996; Hebeisen et al., 1997a; Zanetti et al., 1997; Navas et al., 1999). Schenk et al. (1997a) concluded this was related to a lack of dependence on soil N supply in the legume. The lack of significant change in legume content in our study supports the conclusion that N limitation, if at all, was not severe.

Warming at ambient CO2 concentration did not alter the cumulative herbage yield of the mixture (−3%) although the composition of the mixture was radically altered. In that treatment, clover production was dramatically reduced (55%), while phalaris grew extremely competitively, producing double the biomass of the field temperature, ambient CO2 treatment (Table 1). Navas et al. (1999) showed that the response of species in mixture could be related to their response in monocultures. This is supported in the present study by our observation of poor growth of the clover in the monoculture and the mixture under warming at ambient CO2. Warming had a larger effect on plant bases and root biomass in the mixture. However, responses were intermediate between clover and phalaris monocultures (Table 3), again suggesting the mixture response could be predicted from the monocultures.

The combination treatment of warming and elevated CO2 did not significantly increase phalaris or clover herbage in monoculture. Cut herbage yield of the mixture, however, was significantly increased, in both phalaris and clover components (Table 1), although the proportion of each species was not altered (Fig. 8). In a study of ryegrass pasture turves, yields of herbage increased in winter and spring under elevated CO2 plus a 6°C warming, but this large temperature increase reduced yields in summer (Ross et al., 1995). Ferris et al. (1996) found that shoot growth of perennial ryegrass decreased under elevated air temperature (4°C), with this reduction being greatest in summer. In southern Australian pastures, water limitation during summer results in summer dormancy of phalaris. Therefore negative impacts of extreme summer temperatures would be minimal. Nevertheless, a longer-term study may have shown negative effects of temperature on established phalaris plants.

Overall the study shows that subterranean clover was more responsive to atmospheric temperature and CO2 increase than was phalaris, and it appears that warming may negate the stimulatory effect of elevated CO2 on its growth. The degree of warming experienced under future climate change in south-eastern Australia will affect responsiveness of pastures to the increasing atmospheric CO2 concentration. From this experiment and colleagues, it seems unlikely that the proportion of clover in pastures will diminish under climate change, and this may be a benefit of long-term persistence of subterranean clover, although climate change may also impact on other factors affecting clover persistence such as allelopathy, disease, weed invasion, soil acidification and seedling recruitment. Equally the time scale of discernible climate change is slower than the timescale of plant improvement by breeding, so cultivars are likely to become more adapted to a changing environment in the normal course of plant breeding. Further studies are required to assess the likely impacts of climate change on species composition and persistence in the long term. It appears that climate change is likely to increase herbage yields of subterranean clover-phalaris pastures of southern Australia. However, animal production will be affected by the quality as well as the productivity of pastures (Lilley et al., 2001).


We thank many people who made the study possible: Gus Ingram, Helen Adams, Bruce Reid, Bruce Robertson and Steve Speer for technical assistance; Nola McFarlane, who assisted with construction and testing of the tunnels and photosynthesis measurements; Damian Barrett also assisted with photosynthesis measurements; and we thank John Evans for the use of photosynthesis equipment. Bruce Condon designed and serviced the electronic equipment that ran the tunnels and staff at Ginninderra farm assisted in preparation and maintenance of the site. This work contributed to the CSIRO Climate Change Research Program and was supported by the National Greenhouse Research Program.