The effects of prolonged exposure to elevated temperatures and elevated CO2 levels on the growth, yield and dry matter partitioning of field-sown meadow fescue

Selostus: Kohotetun lampotilan ja kohotetun CO2-pitoisuuden vaikutukset peltoon kylvetyn nurminadan kasvuun, satoon ja kuiva-aineen jakautumiseen

in the dry matter production of C 3 plants, an effect that has already been observed in many experiments (Lawlor andMitchell 1991, Bowes 1993).
Increases in photosynthesis are seldom translated into equivalent increases in crop yields, however, because of other constraining factors in the environment or genetic factors at plant level, which limit plant growth (Nijs et al. 1989, Lawlor and Mitchell 1991, Ryle et al. 1992, Bowes 1993, Gay and Hauck 1994. Alternatively, the extra photosynthate may be translocated to roots (Nijs et al. 1988, 1989, Bowes, 1993, Luoetal. 1994, Schenk et al. 1995 or extracted to the soil as carbohydrate (Diazet al. 1993, Zak et al. 1993. A more extensive root system would be beneficial for grass and for cereal crops through the improvement in nutrient and water availability, and also as a carbohydrate storage for regrowth of grass during the first days after cuts (Nijs et al. 1988). Moreover, increased excretion of carbohydrate through roots to the soil might increase the microbial biomass of the soil, thus either improving (Zak et al. 1993) or diminishing nitrogen availability (Diaz et al. 1993), probably depending on the soil type and the chemical composition (the C:N ratio) of the litter being digested by the microbes.
Increases in concentrations ofCO, along with changes in other atmospheric constituents are also known to affect the global climate. Recent projections for Finland indicate a mean annual warming ofbetween 1.1 and 6.6°C by 2100, with a central estimate of 4.4°C (Carter 1996). Higher growing-season temperatures could be expected to enhance the yield of crops with continuous growth, such as grass, mainly through a lengthening of the growing season. Gains in yield might be especially great in the spring, when the light intensity is high but temperatures are now too low for crop growth.
The aim of the investigation described here was to evaluate, through direct experimentation, the impacts of future climate changes on the growth and yield of a grass crop, meadow fescue (Festuca pratensis). Although many experiments have been conducted to study the effects of C0 2 on individual or several grass plants in pots (e.g. Nijs et al. 1988, 1989, Ryle et al. 1992, Baxter et al. 1994,few investigations have been made of the effects of both C0 2 and temperature on crop stands, especially under the long photoperiod conditions prevailing during the growing season at high latitudes. As the sward was sown directly on the field at normal sward density, the arrangement of the experiments simulated natural conditions closely with respect to canopy structure and intraspecific competition, and there was no pot size limitation for root growth, as may have been the case in many previous experiments (Arp 1991, Thomas andStrain 1991).

Material and methods
The experiments were carried out in southern Finland, Jokioinen (60°49'N, 23°30'E). Meadow fescue (Festuca pratensis Hudson) cv. Kalevi was grown according to the following four treatments: a) ambient air temperature and ambient C0 2 concentration; b) increased temperature (3°C above ambient) and ambient C0 2 ; c) ambient temperature and elevated CO, (700 ppm), and d) increased temperature and elevated C0 2 . To maintain the experimental temperatures at a constant 3°C higher than the ambient temperatures, a greenhouse (20 m x 30 m) was built over an experimental field (Hakala et al. 1996).The experimental field outside the greenhouse, at ambient temperature, was covered at a height of 3-4 m with the same plastic film that was used in the construction of the greenhouse to achieve radiation conditions comparable to those in the greenhouse and to protect the area from rain. The temperatures in the open field under the film cover were recorded, and the greenhouse temperatures were regulated so that they were constantly 3°C higher than these. The mean temperatures during the day (from 6 a.m to 6 p.m.) under the cover in the open field were from I°C lower to I.5°C higher than those in Vol. 5 (1996): 285-298. the adjacent field without a cover (average temperature difference from I June to 31 August O.O°C, standard deviation 0.62°C) . The cover over the open field was removed during the winter to allow snow to protect the grass naturally from freezing temperatures. In the greenhouse the grass crop was covered with a light cloth for periods of extreme low temperatures in winter.
The C0 2 experiments were conducted in open-top chambers (OTCs) 3 m in diameter and 2 m high (Hakala et al. 1996). The OTCs were divided into two halves, the southern half being occupied by the meadow fescue stand, and the other half being used for experiments with spring wheat. Two experiments were conducted, each covering two growing seasons: those of the year of sowing and those of the following year, which is considered as the main yield producing sea-son. Figure 1 shows the temperature regimes of the four growing seasons 1992-1995.
Experiment 1 was conducted on a heavy clay soil. Before the experiments the soil was mixed with 1000 rrvVha of peat containing 35% sand. The peat was limed and fertilized with a peat fertilizer containing 11% nitrogen, 11% phosphorus, 20% potassium, 6.2% sulphur, 1% copper, 0.6% sodium, 0.6% chloride, 0.5% iron, 0.4% manganese, 0.2% zinc, 0.1% molybdenium, 0.1 % magnesium and 0.08% boron. The peat-sand mixture was mixed with the clay soil at a depth of 25-30 cm, after 500 kg/ha ofa fertilizer containing 20% N, 4% P and 8% K had been added. The soil was treated similarly both in the greenhouse and in the open field. At the end of 1993, when experiment 1 was finished, the clay-peat soil of the experimental site was  Table 1). The beginning of the thermal growing seasons is marked with * (for exact dates, see Table 1). s 1 = sowing in the greenhouse, s 2 = sowing in the open field. *2 = beginning of growing season in the open field in 1995. The temperatures inside the greenhouse were 3"C higher than those in the open field (see Hakala et al. 1996 for details). replaced with a lighter sandy loam soil brought to the experimental plots from another field in Jokioinen, and the drainage of the soil was improved by placing new drainage pipes under the soil at a depth of about 40 cm. Because nutrient analysis of the soil showed a deficiency in Ca and Mg, the soil was limed with 8000 kg/ha of a fertilizer containing 35% calcium and 3% magnesium. Before sowing began, 500 kg/ha (greenhouse) or 550 kg/ha (open field) of a fertilizer containing 20% N, 6% P and 6% K was added to the soil surface. With the different fertilization rates, the nitrogen levels inside the greenhouse and in the open field were adjusted to about 120 kg N/ha. The experimental plots were sown directly on the field at 1250 germinating seeds/m 2 in 1992 (experiment 1) and 750 germinating seeds/m 2 in 1994 (experiment 2), row width 12.5 cm. Inside the greenhouse, at elevated temperatures, sowing was completed shortly after the thermal growing season had started, i.e. when the temperature could be expected to stay constantly above 5°C. The sowing dates inside the greenhouse were thus 29 April in 1992 and 14 April in 1994. In the open field there was a delay in sowing, due partly to the later onset of the thermal growing season and partly to the need for the soil to dry sufficiently for sowing. Sowing thus proceeded some 2-3 weeks later in the outside field than in the greenhouse (14 May in 1992 and 9 May in 1994, Table 1).
After emergence, the seedlings were enclosed in OTCs: four in the greenhouse and four in the open field. Two of the four OTCs were maintained at elevated C0 2 levels and the other two at ambient levels, giving two independent replicates per treatment (Hakala et al. 1996). The CO, levels in the high-CO, OTCs were not elevated during the winter, but the C0 2 treatments of the grass were started in April, as soon as the growing season began. Irrigation was applied using drip-irrigation in experiment I and manually in experiment 2, to ensure that soil moisture levels would not limit growth. Fertilizer was administered both in the open field and in the greenhouse at such a rate that the N level of the soil was 120 kg before sowing and in the spring. Thereafter in the first growing seasons the grass was fertilized with 80 kg N after the first cut and with 60 kg N after the second cut both in the open field and in the greenhouse, and with 60 kg N after the third cut but only in the greenhouse. In the second growing seasons the grass was fertilized with 80 kg N after the first cut and 40 kg N after the second and third cuts both in the open field and in the greenhouse.The fertilizer used contained 20% N. 6% P, 6% K, 5% Ca, 2.1 % S, 0.5% Mg, 0.03% B, 0.1% Fe, 0.1% Na and 0.016% Se. Pest and disease control were applied when needed.
In 1992-1994, meadow fescue was cut at approximately monthly intervals (Table 1). During the second growing seasons it was cut for the first time shortly after it began to flower. This occurred in both experiments 2 weeks earlier at elevated temperatures than at ambient temperatures (Table 1). In 1995, following the first cut at the beginning of flowering, the grass was always cut after the leaf area index (LAI) of a stand had reached a value of 5, as measured with an automatic LAI meter (Licor, USA). This resulted in a different number of cuts in each treatment (Table 1). Biomass samples were collected during the cuts from 10randomly chosen sampling plots in each of the chambers, each sampling plot comprising 15 cm of planted row. For biomass, nitrogen and yield determinations, the samples were oven-dried (2 hours at 100°C, then overnight at 60°C). The samples for soluble carbohydrate measurements were freeze-dried and stored at -20°C before analysis.
Root samples were taken at a depth of 20 cm during the first and second cuts in the first growing season of experiment 1 (1992), and during the last cuts in the other growing seasons (1993)(1994)(1995). The sample size was 15 cm of row, and both tiller number and root biomass were determined from the samples.
The nitrogen content of the above-ground biomass was determined in 1992 with the Kjeldahl method with a Kjeltec System 1026 Distilling Unit (Tecator AB, Sweden), and with CuS0 4 *SH,O as the catalyst. For details of the procedure, see manual of the Kjeltec System 1026 Distilling Unit. In 1993, the nitrogen levels were not measured because insufficient resources were available for the laborious Kjeldahl procedure. In 1994-1995,the nitrogen content was determined with an automatic nitrogen analyser, LECO FP-428 (LECO corp., USA). For the soluble carbohydrate analysis, the plant samples were first inverted overnight at +5O°C in 0.1 N HCI. They were then filtered, and the filtrate was mixed with ion-exchangers, shaken for one hour and refiltered; 2-ml samples of this filtrate were then used for the analysis of soluble sugars (glucose, fructose, saccharose and fructosans). The analysis was performed photometrically (Schimadzu, Japan, 540 nm), according to the method of Nelson (1944) and Somogyi (1945), but modified in such a manner that after the addition of copper reagent the test tubes were filled with distilled water to 7 ml before the arsenomolybdate reagent was added.
To give an idea of the variation between samples, standard errors of the mean were calculated over all the samples taken from each treatment at each sampling time. The "n" in the tables and figures refers to the total number of samples of both replicates.

Number of tillers and yield
The number of tillers was counted in connection with the first and second cuts in the first growing season of experiment 1 (1992) and in connection with the last cuts in all the other growing seasons (1993)(1994)(1995). The number of tillers/15 cm of row was 1.7-2.7 times higher (depending on the treatment) in the first grow-ing season and 2.0-2.2 times higher in the second growing season of experiment 2 than in the first and second growing seasons of experiment 1 (Fig. 2). There was no consistent effect of the temperature treatment on the tiller number of meadow fescue, but the increase in C0 2 concentration increased the tiller number by 63% and 90% at ambient temperatures and 20% and 55% at elevated temperatures during the first growing seasons of experiments I and 2, respectively (counted on 14 July-14 August 1992 and 30 September 1994) (Fig. 2). The tiller number also increased in elevated C0 2 during the second growing seasons of experiments 1 and 2 at ambient temperatures (by 24% and 14%, respectively), but no increases were recorded at elevated temperatures (Fig. 2). The total above-ground yield (cumulative agricultural yield of the growing season, excluding stubble biomass) of meadow fescue was 32% (experiment 1) and 41% (experiment 2) higher at elevated temperatures in ambient CO, than at ambient temperatures and ambient C0 2 in the first growing seasons (Fig. 3). During the second growing season of experiment 1, when there were four cuts both inside the greenhouse and in the open field, the yield was 15% higher in the greenhouse in ambient CO, than in the open field. As the crop was cut whenever it attained a LAI of 5 during the second growing season of experiment 2, there were six cuts inside the greenhouse in ambient CO,, and four cuts on the equivalent plots in the open field, and the total above-ground yield of the plots inside the greenhouse was 65% higher than in the open field (Fig. 3).
The total yield in a growing season was about 10% higher in elevated than in ambient C0 2 at both ambient and elevated temperatures in the first growing season of experiment 1. In the second growing season, the total yield was still 10% higher in elevated C0 2 at elevated temperatures, but there was no difference in total cumulative yield at ambient temperatures (Fig. 3). In experiment 2 the yield of meadow fescue was 29% (first growing season) and 22% (second growing season) higher in elevated than in ambient C0 2 at elevated temperatures, but more or less the same in both CO, treatments at ambient temperatures (Fig. 3).
In experiment 1, at elevated temperatures, the increase in the total yield in elevated CO, was achieved before the first cut, after which CO, enrichment had no effect on growth rate (Fig. 3). At ambient temperatures, an increase in yield in elevated C0 2 also occurred early in the growing season, i.e. between the first and the second cuts during the first growing season and before the first cut during the second growing season. During the second growing season, however, the growth rate in ambient C0 2 increased markedly relative to that in elevated C0 2 after the second cut, resulting in an equal total yield in both C0 2 treatments. In experiment 2, at elevated temperatures, the favourable effect of C0 2 enrichment on growth rate was evident until late in the season (August), after which the effect was slight (first growing season) or negative (second growing season). At ambient temperatures the growth rate was more or less the same in both ambient and elevated CO, (Fig. 3).

Root biomass
Root biomass was measured both in the open field and in the greenhouse at the end of the 1993-1995 growing seasons, and during the second cut in 1992 (Fig. 4). Root biomass increased markedly towards the end of both experiments.
It was approximately the same in ambient C0 2 treatments both at ambient and at elevated temperatures except in the second growing season of experiment 1, when the root dry weight was higher at ambient temperatures (Fig. 4). At ambient temperatures, the root dry weight was 67% (first growing season) and 38% (second growing season) higher in elevated C0 2 in experiment 1 and over 30% higher in experiment Fig. 3. Cumulative yield (g dry weight of above-ground biomass/15 cm of planted row) during the growing seasons of experiment 1 (1992)(1993) and experiment 2 (1994)(1995). The numbers on the lines represent the daily growth rate of the grass during the period between cuts. The number of samples (n) was 17-20 in all cuts except at ambient temperatures in 1993, when it was 7, and at ambient temperatures in 1995, when it was 14. The standard error of the mean of the biomass samples at different cuts was in general around or less than 10% of the mean, but in 1993, at ambient temperatures, it averaged 19% of the mean. weight, being generally higher in experiment 2 than in experiment 1 (Table 2), possibly because of the different soil types. The nitrogen content was 5-22% higher in elevated C0 2 than in ambient CO, in experiment 1 (measured only in the first growing season) at both temperatures. In experiment 2 the nitrogen content of the aboveground biomass of the grass was 2-17% higher in elevated C0 2 than in ambient C0 2 at ambient temperatures, and 3-12% lower at elevated temperatures (Table 2).

Soluble carbohydrate content of shoots and roots
The soluble carbohydrate content of meadow fescue above-ground biomass was 5-48% higher in elevated than in ambient C0 2 most of the measuring times during summer, but there was no effect of elevated C0 2 on the carbohydrate content in the autumn (Table 3). The effect of the elevation of CO, levels on the carbohydrate content of the roots was inconsistent; in the first growing season of experiment 2 the carbohydrate content of the roots was lower in elevated C0 2 at both temperatures, but in the second growing season it was the same at ambient temperatures and higher at elevated temperatures in elevated CO, (Table 3). Fig. 4). At elevated temperatures, the rise in CO, did not increase root weight in experiment 1; in experiment 2, however, it increased root weight by 29% (first growing season) and 8% (second growing season) (Fig. 4).

Nitrogen content of above-ground biomass
The nitrogen content of meadow fescue aboveground biomass ranged from 2% to 5% of dry

Discussion
The total cumulative agricultural yield of meadow fescue, cv. Kalevi during the growing season was high compared to the long-term average (1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995) of the official variety trials of the same cultivar from the same area (Järvi et al. 1996). The yield of the stands at ambient temperatures and ambient CO, levels was already 10% higher in the years of sowing (1992 and 1994) than the long-term average for cv. Kalevi achieved in variety trials in the same agricultural area. In the year after sowing, the yield was  from 63% (in 1993) to 105% (in 1995) higher than the average yield of the grass in the variety trials.
The total above-ground yield ofmeadow fescue was higher at elevated than at ambient tem-peratures in all the growing seasons studied. In the first growing seasons, the higher total yields at elevated temperatures were partly due to the longer growing period, since the grass was sown 2-3 weeks earlier in the greenhouse than in the Table 3. Soluble carbohydrate content (% of dry weight) of meadow fescue above-ground (shoots) and below-ground (roots) biomass in 1992-1995. n  11.9 (±0.6) n=2 21.9.1994 9.4(±0.5)n=3 9.1 (±1.1) n=3 12.9.1994 9.3 (±0.2) n=6 9.6 (±0.8) n=6 6.6. open field, but higher growth rates at elevated temperatures also contributed to the increase in yield, especially in 1994 (Fig. 3). In the second growing season of experiment 1, the yield in the greenhouse was only 15% higher than that in the open field. Because the grass was not cut according to LAI, but at about monthly intervals during the second growing season of experiment 1, the small difference between yields in the open field and inside the greenhouse may have been caused by stronger self-shading and, consequently, a lower growth rate of the grass in the greenhouse. When the crop was cut each time it attained a LAI of 5 (in 1995), the total yield in the greenhouse was about twice that in 1993, and the growth rate between the cuts was clearly higher than in 1993, resulting in 65% higher yields than at ambient temperatures. Although favourable, the effect of C0 2 on the yield of meadow fescue was less favourable than expected. Earlier investigations and reviews have reported increases in plant biomass of between 20% and 209% at CO, levels double those of the ambient level (Lawlor andMitchell 1991,Bowes 1993). Even though the photosynthetic rate of meadow fescue in the present experiment was 40-60% higher in elevated than in ambient C0 2 at both ambient and elevated temperatures (Heliö et al. 1995), the increase in photosynthesis was not translated into an equal increase in biomass. Similar results have been reported previously with grasses, and the lack of a greater effect of elevated C0 2 on the yield has been attributed to low sink strength due to limited tiller formation capacity (Ryle et al. 1992, Gay and Hauck 1994. Schenk et al. 1995. After experiment 1, control of C0 2 levels in the C0 2 -enriched OTCs was improved; the claypeat soil was replaced with sandy loam, a soil with good moisture properties, and the drainage of the soil was improved. Moreover, the sowing rate was lowered from 1250 to 750 germinating seeds/m 2 , and cuts were adjusted to take place according to the LAI of the grass (in 1995). These changes were made to promote the effect of C0 2 on the tiller number and yield, as it is known that tiller development and yield are en-hanced in soils with good nutrient availability, and when the light penetration into the canopy is improved (by lowering the sowing rate and by more frequent cuts) (Williams 1982). The tiller number/sample increased markedly in all treatments after these changes were made, and the effect of elevated CO, on the number of tillers was greater in the first growing season of experiment 2 than in the first growing season of experiment 1. However, in the course of experiment 2, just as in experiment 1, the effect of elevated C0 2 on the tiller number declined, being even smaller at the end of experiment 2 than at the end of experiment 1.
Because the tiller number was approximately 2-3 times higher in each treatment during experiment 2 than during experiment 1, there should have been enough room for the tillers to develop. Thus canopy closure cannot explain the lack of effect of elevated C0 2 on the tiller number at elevated temperatures in the second growing season of experiment 1. In the second growing season of experiment 2, in contrast, canopy closure may have been responsible for the decline in the effect of C0 2 on tiller development.
Even when the tiller number was greater in elevated than in ambient C0 2 (Fig. 2), the total yield per cut did not necessarily increase (Fig. 3). The failure to produce higher yields in elevated C0 2 , even if the tiller number increases, has been observed with Phleum pratense and Lolium perenne (Saebo and Mortensen 1995), Agrostis capillaris, Dactylis glomerata and Festuca pratensis (Saebo and Mortensen 1996) and has been attributed to the lower dry weight of individual tillers, due at least partly to reduced plant height. At elevated temperatures, the tiller number was the same in the CO, treatments at the end of the second growing seasons in both experiments, even if the yield was higher in elevated C0 2 (Figs 2 and 3). As the tiller number was counted only at the end of the second growing seasons, the gain in biomass may have been due to an increase in the tiller number earlier in the season, as suggested by the increase in the yield in elevated C0 2 that took place early in Vol. 5 (1996): 285-298. the season (before the first cut) in 1993, and by the reduced daily growth rate in elevated C0 2 relative to that in ambient C0 2 late in the season in 1995.
Closure of the plant canopy leads to poorer light penetration and hence lower biomass production than expected on the basis of higher assimilation rates in elevated C0 2 measured at saturating light (Du Cloux et al. 1987, Nijs et al. 1989, Schenk et al. 1995. When the grass was cut during the second growing season of experiment 2 (1995) according to the LAI of the individual replicates, canopy self-shading was reduced and made more or less similar in all treatments. The higher yields both in the open field (29% greater) and in the greenhouse (85% greater) in ambient CO, in 1995 than in 1993 (the second growing season of experiment 1) (Fig.  3) may have been partly due to reduced selfshading and better use of light for photosynthesis and growth at canopy level, although the effects of different soil type and different climatic conditions should not be discounted. Nevertheless, even with conditions apparently favouring a positive effect of enriched CO, on yield in experiment 2, such an effect was still only observed in the greenhouse.
In cases when the above-ground biomass does not increase in proportion to the increase in photosynthesis, there have been reports of the extra photosynthate being translocated to the roots (Nijs et al. 1988, Luo et al. 1994, Schenk et al. 1995 or to the stubble (Saebo and Mortensen 1995). There was evidence of this occurring here, too, as the root fraction increased in elevated C0 2 even though the elevation ofC0 2 had no effect on the above-ground biomass at ambient temperatures in experiment 2 (Figs 3 and 4). As there was no decrease in the nitrogen content in the shoots of meadow fescue at ambient temperatures in either experiment 1 or 2, it is improbable that the root fraction would have increased mainly to make more of the nitrogen and other nutrients available for the plant, as proposed in another study on grass swards (Schenk et al. 1995), but the response seems to be rather a way of translocating extra photosyn-thate to roots when the above-ground biomass cannot use it efficiently enough (Luo et al. 1994). Even if the yield of the grass did not increase at ambient temperatures in elevated C0 2 in experiment 2, the increase in root biomass could have been beneficial for grass growth during drought or if nutrients had become scarce, and could also have served as a bigger carbohydrate reserve for the grass to help it survive the winter better.
Accumulation of carbohydrate in the leaves and/or an increase in the number of mesophyll cell layers have been interpreted as another sign of the reduced sink strength (Ryle et al. 1992, Bowes 1993,Baxter etal. 1994, Luo et al. 1994 often associated with a decrease in the nitrogen content of the leaves (Campbell et al. 1988, Ryle etal. 1992,Bowes 1993,Luo et al. 1994, Schenk et al. 1995. The decrease in the nitrogen level in the leaves has been explained, at least partly, by dilution of nitrogen with the carbohydrates (Campbell et al. 1988, Luo et al. 1994 or by translocation of the nitrogen from rubisco and photorespiratory enzymes to other proteins more limiting to photosynthesis, or to other parts of the plant to increase the sink strength in elevated CO, (Hocking and Meyer 1991, Stitt 1991, Conroy and Hocking 1993. The carbohydrate content of the leaves was not measured separately here, but that of the total above-ground biomass was higher in elevated CO, at both ambient and elevated temperatures at most of the measuring times during summer (Table 3).
According to the official variety trials conducted in 1988-1995, the average protein content of the above-ground biomass of meadow fescue cv. Kalevi is 15%, or 2.3% N of dry weight (Järvi et al. 1996). The nitrogen levels in meadow fescue in the present experiments were thus approximately average or slightly above average in 1992, and clearly above average in 1994-1995. Despite the increase in the carbohydrate content of the above-ground biomass in elevated C0 2 , the nitrogen content of the aboveground biomass was not reduced in elevated C0 2 , except for a slight decrease at elevated temperatures in experiment 2. Rather, there seemed to be an increase in the uptake of nitrogen in ele-vated CO, at ambient temperatures in all the years that nitrogen content was studied, and at elevated temperatures in experiment 1 (in 1992).
It has been noted that the effect of elevated C0 2 on photosynthesis is lower at low light intensity (Gay and Hauck 1994) and may even be reduced relative to that in ambient C0 2 (Nijs et al. 1989). Lowered photosynthesis rates at low light intensities, e.g. inside the grass canopy or on cloudy days, would not have been detected here, because the measurements reported by Heliö et al. (1995) were always performed at saturating light intensity. The measurements were also usually conducted in the morning or early afternoon; thus any late-afternoon decreases in the photosynthetic rate (expected to take place because of sink size limitation and the consequent accumulation ofphotosynthetic products in the leaves) would not have been detected (Nijs et al. 1989). It is possible that the relatively minor effect of CO, on the yield of meadow fescue, especially at ambient temperatures, was partly due to lower photosynthetic activity of meadow fescue in elevated than in ambient CO, at non-saturating light intensities and/or during the late afternoon if both the sink capacity and the capacity of the leaves to store carbohydrate were restricted. The increases in the dry weight of the root fraction at ambient temperatures without any increase in the above-ground biomass also suggests the restricted sink capacity of the above-ground biomass (Luo et al. 1994). Evidence further exists for significantly higher dark respiration rates in elevated CO r causing a large part of the C0 2 assimilated during the day to be released during the night (Nijs et al. 1989, Wolfenden andDiggle 1995). This possible mechanism suppressing biomass increases in enriched CO, remains to be investigated. Diurnal changes in the photosynthetic activity of meadow fescue in elevated and ambient CO, concentrations will also be the subject of future experiments.