• Biomass of genotypes of Carex flacca and Bromus erectus under CO2 enrichment was measured to test for ‘winners’ and ‘losers’ within the populations. Allocation was studied to identify reasons for varying biomass responses of the genotypes.
• In a first experiment, genotypes of both species were grown separately and mown five times. In a second experiment, introducing species interaction, species were grown together (using nine genotypes of each) and harvested destructively during that time.
• The first experiment showed that growth stimulation across harvests (Bromus+15%; Carex+11%) and genotype × CO2 treatment interactions in single harvests was significant, but repeated measures ANOVA was not (Carex) or only marginally significant (Bromus). Destructive harvests in the competition experiment indicated a significant growth stimulation (Bromus+28%, Carex+35%), but no CO2 × genotype interaction was found. Similarly, ANOVA of dry matter allocation never indicated a significant CO2 × genotype interaction.
• The lack of a sustained CO2 × genotype interaction in isolated and competitive growth conditions with good moisture supply suggests, that if there is such an interaction, it must be very small.
Interspecific and intraspecific variation of fitness related traits is critical in determining a species’ and genotypes’ potential to evolve and persist under altered environmental conditions, such as rising ambient CO2 concentration (Woodward et al., 1991; Körner et al., 1996). Thus, differences in the CO2-responsiveness of species within a plant community and the responsiveness of genotypes within a population may determine future genetic composition of biomes. To date, research has indicated distinct differences in the responsiveness of plant species to elevated atmospheric CO2 concentration (Warwick et al., 1998; Volk et al., 2000). In our reference calcareous grassland, the subdominant Carex flacca and the dominant Bromuserectus showed biomass increases of 249, 33% respectively (third season of CO2 enrichment; Leadley et al., 1999). The average biomass response to CO2 enrichment for wild herbaceous plants in isolation was reported as +35% (62 species; Poorter, 1993) and +44% for nondomesticated Poaceae (meta-analysis by Wand et al., 1999). Genotypic differences in the CO2-responsiveness of growth reflect a much less consistent picture. Some work has shown no significant CO2 × genotype interaction (Fajer et al., 1992; Curtis et al., 1994; Schmid et al., 1996; Lüscher et al., 1998; Roumet et al., 1999) while others have (Leadley & Stöcklin, 1996, glasshouse test; Stewart & Potvin, 1996; Steinger et al., 1997, field test). This varied response has led us to question whether differentiation among species or genotypes would separate future ‘winners’ from ‘losers’, and if so, why. Patterns of dry matter allocation together with leaf properties (specific leaf area, photosynthesis) may permit an explanation (functional growth analysis; Poorter & van der Werf, 1998). We hypothesized that inter- and intraspecific differences in the biomass response of two important grassland graminoids do exist, and that these responses will be reflected in CO2× allocation interaction. We also assumed that an evolutionary effective difference in the biomass response of genotypes would require a persistent CO2 treatment × genotype interaction over sequential growth phases.
In order to understand such CO2× biodiversity phenomena, we set up two multiple harvest experiments under nonmoisture limited indoor growth conditions. In the first experiment, clones of Carex flacca and Bromus erectus were grown in isolation and mown five times. In the second experiment, the two species were planted together in large containers, simulating a natural situation. Relative growth rate (RGR), relative biomass response, the rate of photosynthesis, specific leaf area (SLA), specific root length (SRL) and leaf and root mass fractions (LMF, RMF) were measured during the 4 month growth period. In order to distinguish plant size induced changes in biomass allocation from CO2 induced changes in allocation (see Coleman et al., 1994) an allometric coefficient (k) was computed, which permits testing for any deviation from the control’s dry matter fractionation pattern.
Materials and Methods
Experimental plants were ramets of 14 genets of Bromus erectus (Huds.) and of 9 genets of Carex flacca (Schreb). In March 1996 parent plants were collected at our reference field site, covering the whole site in order to provide a good representation of available genotypes. Bromus is the dominant species and contributes almost half of total plant biomass (Huovinen-Hufschmid & Körner, 1998). Carex is a subdominant, rhizomatous sedge, typical for these periodically dry and nutrient poor grasslands.
Experimental design and treatment
For the first experiment (plants growing in isolation; five harvests), a total of 36 ramets of Carex and 56 ramets of Bromus were planted in natural soil in pots of 0.6 l each. Very small and large ramets were disposed, so that there were no significant differences between individuals of clone- or treatment-groups (f. wt). Pots were distributed to four naturally lit, air-conditioned glasshouse tents with temperature maintained close to outdoor conditions and two supplementary 1000-W daylight halogen lamps per glasshouse. Two glasshouses each were maintained at 365 µl l−1 and 600 µl l−1 CO2, with infrared photo-acoustic gas analysers (CO2-Controller 7MB1300, Siemens, München, Germany) and mass flow controllers. At the first harvest (15 wk after planting) and the second harvest (24 wk after planting) Carex was cut at 4 cm and Bromus at 7 cm above ground level. After the second harvest, experimental plants were thinned out for clonal propagation (while retaining treatment affiliation) and were replanted to new soil. Pots were then moved into two 3 m2, fully air-conditioned growth chambers (Weiss Umwelttechnik GmbH, Reiskirchen, Germany) for the simulation of another growing season and regrowth was cut after 13, 20 and 25 wk as described above (harvest 3–5). One chamber was maintained at 365 µl l−1, the other at 600 µl l−1 CO2 using the equipment described above. Illumination was provided by high-pressure discharge lamps (MF 400 LE/BUH, EYE, Tokyo, Japan). Photosynthetically active photon flux density (PPFD) of 690 µmol m−2 s−1 at canopy level was supplied for 10- to 12-h, plus 1 h of 350 µmol m−2 s−1 PPFD at ‘dusk’ and at ‘dawn’, yielding a seasonally adjusted 12–14 h daylight period. Temperature was 18° : 10°C (day : night). Relative humidity was 60% : 80% (day : night). Plants were watered regularly with deionized water. No nutrients were added.
For the second experiment (introducing species interaction; three harvests), 162 plant assemblages were accommodated in the same two growth chambers at the [CO2] as described above for a simulated 4-month season (132 days). Light was supplied over a 12- to 17-h, seasonally adjusted daylight period. Temperature was 20° : 12°C (day : night) until day 68 and 24° : 14°C thereafter. Experimental plants were ramets of nine genets of Bromus and of Carex from the genotypes used in the first experiment, cloned under ambient [CO2] until > 486 rooted tillers for planting were available. The plant assemblages formed randomly chosen pairs of genotypes of the species. Two ramets of one genet of Carex and Bromus, that is a total of four evenly spaced ramets were grown in round, 50 cm high × 16 cm wide (9 l), polyethylene tubes with a drainage mat and holes at the bottom. The upper 10 cm of the substrate (main rooting horizon) consisted of one part natural soil from the field and two parts of calcareous marl and sand to facilitate later root washing and to compensate for the increase of soil fertility following disturbance (excavation, sieving). The lower 38 cm consisted of a 50 : 50 mixture of calcareous marl and sand. This nutrient-poor combination of substrates resembles the natural growth situation. Water was supplied equally to all plant-containers twice a week and provided nonlimiting soil water content for all plants throughout the trial. After 6, 9 and 19 wk of growth, 54 plant assemblages were harvested. Replication on plant assemblage level was n = 3 for genotype, n = 27 for treatment and n = 54 for harvest.
Measurements and data analysis
In the first experiment, all mown plant material from repeated above-ground harvests was dried at 80°C for 36 h and weighed.
In the second experiment complete plant assemblages were washed free of soil and marl on sieves. While floating in water, plants were separated and sorted by species and clone. After cool storage (at c. 1°C up to 24 h) plants were divided into green leaves, attached dead leaves, leaf sheaths, roots and rhizomes (the latter in Carex only). No litter was found. The length of main roots was determined for the calculation of specific root length. Green leaf area was measured with a photo-planimeter (Li-Cor 3100 Area Meter, Lincoln, NE, USA) for the calculation of SLA. Plant material was then dried at 80°C for 36 h and weighed. Mean RGR between two consecutive harvests (‘1’ and ‘2’) was calculated as (ln : d.wt2– ln : d.wt1)/(time2– time1) (Causton & Venus, 1981).
The rate of change of dry matter fractions in relation to total plant d. wt, was expressed by calculating the allometric coefficient (k), which represents the slope of the linear regression of log transformed biomass components vs log transformed total d. wt. Allometric coefficients were calculated for the relationships of leaf mass, senesced leaf mass (Bromus), root length, root mass, fine root mass and total root mass vs total plant mass. A higher k under CO2 enrichment indicates the biomass fraction to increase at a faster rate than the total biomass compared to ambient [CO2]. Thus, confounding of size effects with treatment effects is avoided. The maximum rate of photosynthesis was assessed between the 2nd and 3rd harvest at 1000 µmol m−2 s−1 PPFD and at the growth CO2 concentration, using a steady state gas exchange system (Li-Cor 6400, Li-Cor).
To minimize unwanted effects of glasshouse (containing four fumigation tents) and growth chamber (two rooms), container position within glasshouse tents or phytotron rooms was randomized twice a week, and container groups with associated CO2-control were moved between the two rooms weekly to avoid pseudoreplication. Single harvest ANOVA and repeated measures ANOVA (RM-ANOVA) across harvests for both experiments was conducted with JMP 3.2.2 (SAS Institute, Cary, NC, USA). We used CO2-treatment, clone and the CO2-treatment × clone interaction as factors. In the first experiment with differing numbers of clone-replicates between harvests, biomass values for each clone × CO2-treatment combination were randomly assigned to two groups, thereby reducing data to n = 2 replicates for each harvest.
Isolated plant test including five above-ground harvests
Performance ranks of genotypes were inconsistent between harvests in both species (Fig. 1). Bromus showed a significant genotype × CO2 interaction in the biomass response of all harvests (except the 2nd), suggesting a clone specific response to CO2 enrichment (Table 1). In Carex the CO2 × genotype interaction was only significant in harvests 3 and 4. But RM-ANOVA across all five harvests revealed no significant CO2 × genotype interaction in the biomass response of Bromus and Carex.
Table 1. ANOVAs for the effect of CO2 and genotype and their interaction on the above-ground biomass of Bromus and Carex in the first experiment. Data were log-transformed before analysis
Relative CO2 response of above-ground biomass in Bromus (Table 2) ranged among harvests from +46% to −8% with a mean across harvests of +15%. In Carex, yield ranged among harvests from +60% to −30%, with a mean across harvests of +11% and surprisingly similar to that in Bromus. In a species comparison, biomass of Carex at the 2nd harvest responded more positively (+28%; P = 0.018) to CO2 than that of Bromus. By contrast, at the 4th harvest, Bromus responded more positively. RM-ANOVA yielded no species × CO2 interaction across harvests in spite of these differences.
Table 2. CO2 effect on above-ground biomass (Carex cut at 4 cm, Bromus cut at 7 cm) in the isolated plant test
Amb. CO2 ± SE
High CO2 ± SE
Amb. CO2 ± SE
High CO2 ± SE
Values are mean d. wt (d. wt in g) of a species per container. SE, ± 1 standard error. Amb. CO2, 365 ppm CO2 treatment. High CO2, 600 ppm CO2 treatment. Resp.%, growth response as ((d. wt high CO2– d. wt amb. CO2)/d. wt amb. CO2) × 100.
0.373 ± 0.020
0.501 ± 0.032
0.381 ± 0.030
0.607 ± 0.056
0.474 ± 0.024
0.458 ± 0.026
0.484 ± 0.026
0.606 ± 0.031
0.043 ± 0.004
0.040 ± 0.003
0.042 ± 0.003
0.030 ± 0.003
0.444 ± 0.022
0.646 ± 0.027
0.479 ± 0.016
0.501 ± 0.019
0.794 ± 0.033
0.828 ± 0.035
0.441 ± 0.014
0.422 ± 0.014
Competition experiment including three whole-plant harvests
Genotype × CO2 interaction
In both species, biomass response (Fig. 2) showed no significant genotype × CO2 interaction at any single harvest (Table 3). Biomass and the allocation indicators studied, also failed to show a significant genotype × CO2 interaction in the RM-ANOVA. Thus, the CO2 response of total plant biomass as well as the response of the shoot parameters SLA, LMF (both species), percentage of senesced leaf area (Bromus only) and of the root parameters SRL and RMF was independent of the genotype.
Table 3. ANOVAs for the effect of CO2 and genotype and their interaction on the whole plant biomass of Bromus and Carex in the second experiment. Data were log-transformed before analysis
Biomass response and relative growth rate (RGR) of species
In the 1st and 2nd harvest the total biomass of both species increased by a factor of four or more (Fig. 2), but no significant CO2 effect was recorded (Table 3). Significant treatment effects became apparent after the 2nd harvest (60 d) and at the final harvest, a +35% stimulation in Carex (P = 0.001) and +28% in Bromus (P = 0.007) was observed. Repeated measures analysis for all three harvests revealed the positive CO2 effect on total biomass to be significant in Carex only (P = 0.014). Neither RM-ANOVA nor ANOVA for individual harvests yielded a significant CO2 × species interaction.
Relative growth rates of both species (Fig. 3) showed a positive CO2 effect (Bromus+28%, P = 0.026; Carex+20%, P = 0.038) over the 3rd interval. These RGRs were significantly different between species (P < 0.001), showing that Bromus was growing substantially faster than Carex, but repeated measures analysis for all harvests indicated no CO2 effect for either species. A CO2× species interaction effect on RGR was also not detected. Interspecific comparison across CO2-treatments and all harvests showed that mean RGR of Bromus was 37% greater than that of Carex.
Leaf mass fraction under CO2 enrichment was significantly lower at 1st harvest in Carex only (P < 0.001), but not in Bromus (Table 4). In the 2nd harvest in Carex the k-value for leaf mass showed a significantly steeper slope in 600 ppm (+0.46) vs 365 ppm CO2 (+0.31; P = 0.018). Bromus showed the same reversal of the overall trend among the 2nd harvest samples, but the increase was weaker under elevated CO2 (slope +0.25 in 600 ppm vs +0.32 in 365 ppm; P = 0.023). The CO2 effect on RMF remained insignificant for both species. Differences in LMF and RMF among harvests by far exceeded CO2 induced changes. Both leaf- and root mass fraction showed no significant CO2 × genotype interaction in the RM-ANOVA of Carex and Bromus.
Table 4. CO2 effect on biomass allocation patterns during three consecutive harvests in the species interaction experiment
Amb. CO2 ± SE
High CO2 ± SE
Amb. CO2 ± SE
High CO2 ± SE
Leaf mass fraction (LMF) and root mass fraction (RMF) values represent whole plant d. wt fractions. Specific leaf area (SLA) values in m2 kg−1, specific root length (SRL) values in m g−1. SE, ± 1 standard error. Amb. CO2, 365 ppm CO2 treatment. High CO2, 600 ppm CO2 treatment. Asterisks indicate P-values: ***(P < 0.001).
Specific leaf area (SLA) and specific root length (SRL)
Analysing the harvests separately revealed a CO2 effect (P < 0.001) in both species at the 3rd harvest (Table 4), but there was no CO2 effect on SLA in the repeated measures analysis for either species.
There was no significant CO2 effect on SRL of Bromus and of Carex in the individual harvests ANOVA. SRL exhibited no CO2 effect in either species in the repeated measures. SRL of Bromus remained stable over time (11.7 m g−1 averaged over harvests and treatments), but SRL of Carex dropped in the 3rd harvest to only 67% (equivalent to 13 m g−1) of its 1st harvest value.
Maximum rate of photosynthesis under growth CO2-concentrations, yielded a significant 23% enhancement (P = 0.004) for Bromus and a 32% enhancement (P = 0.066) for Carex (Fig. 4). These values closely resembled the differences in biomass responses that the species achieved at the third harvest.
By contrast to the foliage of Carex that remained green, senesced leaf area in Bromus reached 30% in ambient and 33% in elevated CO2 at the 3rd harvest, a small, but significant CO2 effect (P = 0.05). Corrected for plant size (k) this effect was retained, hence was directly associated with [CO2]. The RM-ANOVA revealed no CO2 effect.
Genotype × CO2 interaction
Despite CO2 × genotype interactions (Table 1) and variation in performance-rank among the harvests of the isolated plant experiment (Fig. 1), no interactions were found in the harvests of the competition experiment (Table 3). In both experiments genotypic differences in growth were significant, yet RM-ANOVA did not identify a CO2 × genotype interaction indicating sustained response of single clones. This may help explain why some experiments on intraspecific differences in biomass responsiveness found genotype effects and colleagues did not (see Introduction). Steinger et al. (1997), reported CO2 × genotype interaction for Bromus from our reference field site, but the effect was present only when they experimentally excluded competition. Similarly, Wayne & Bazzaz (1995) showed a CO2 × maternal family interaction at low competition, which was reversed at high competition. Thus, nonCO2 environmental interactions specific to developmental stages of the organisms may have determined genotype responses.
Further support comes from the RM-ANOVA of allocation parameters in Bromuserectus and Carex flacca. The critical biometric growth determinants were not affected by CO2 in a genotype-specific manner when studied over an extended period. Most importantly, LMF and SLA which compose leaf area ratio (LAR, m2leaf kg−1plant), identified by Poorter & van der Werf (1998) as the single most important factor determining relative growth rate of plants, did not show a genotype × CO2 interaction.
We suggest that the CO2 × genotype interactions observed in this study reflect a genotype-specific nonCO2 environment and developmental status interaction. The CO2 × genotype interactions from single harvests reported in the literature represent spot measurements in time and have not been successfully tested for consistency over a time-course. Thus, the single harvest response of a genotype to CO2 enrichment seems not sufficient to distinguish future long-term ‘winners’ from ‘losers’.
CO2 responses of species
Biomass and RGR response to CO2-enrichment
The maximum biomass responses observed in the first experiment with isolated species (Bromus+46%, Carex+60%) and in the second test including species interaction (Bromus+28%, Carex+35%), matched the Wand et al. (1999) meta-analysis of CO2 responsiveness in 80 C3 Poaceae species (+44%). Poorter (1993) reported +35% biomass response for 62 wild herbaceous C3 species in high [CO2]. Given our poor soil nutrient conditions, there is a surprising similarity to these studies and a strong contrast to the observations at our reference field site (Leadley et al., 1999). Above-ground biomass of Carex was found to increase by 249% in the 3rd season, while the Bromus biomass increased by only 33% in the 3rd season of CO2 enrichment. Stöcklin et al. (1997, 1998) used intact soil monoliths from the same grassland in a glasshouse experiment and also found strong positive biomass responses in Carex, but negative ones in Bromus. In both of these studies plants experienced periodic drought cycles, as is typical for these calcareous grasslands.
The seeming contradiction with the results presented here can be resolved by the analysis of Volk et al. (2000), who found the biomass CO2 response of Bromus and Carex to be negatively correlated with plant moisture availability. With ample water supply CO2 responses were equally low (+17% in Carex and +19% in Bromus), similar to our +35% in Carex and +28% in Bromus. However, under dry conditions, relative responses to CO2 enrichment were much higher and species specific (Carex+102% and Bromus+42%). From the above we conclude, that within the experimental timescale and under growth conditions of nonlimiting soil moisture a biomass response of c. +30% represents the response potential for both Bromus and Carex. A differential species response seems to depend on additional environmental factors such as periodic drought, during which Carex clearly takes a relative advantage under elevated CO2.
The relative growth rate of both species showed positive as well as negative (nonsignificant) CO2 effects over the course of the 4-month period, similar to the range of responses found by Roumet et al. (1996). The low RGR found here was probably due to the infertile substrate, with the higher RGR of Bromus possibly resulting from its higher SLA (cf. Van der Werf et al., 1993). Driven by the higher growth rate, Bromus eventually yielded a biomass similar to that of Carex, although Bromus tillers had only half the initial biomass of Carex.
Development and size dependent responses constitute a serious complication in the evaluation of treatment responses, including those to elevated CO2 (e.g. Loehle, 1995; Lutze & Gifford, 1998; Stirling et al., 1998). If the question addresses the CO2 effect per se, the effect of size dependent (ontogenetic) change of the allocation pattern has to be distinguished from changes that are directly CO2 induced. The ANOVA of the linear regression between total biomass and one of its components accounts for size dependent biomass fractionation and yielded only few significant CO2 effects. Thus, allocation patterns observed under high CO2 would have been the same in plants of the same size without CO2 enrichment.
Specific root length reduction of Carex roots may reflect maturation, typical for the long-lived roots in this genus. Field data from the reference field site (Leadley et al., 1999) showed the same behaviour for ecosystem SRL. Plants with high SRL are probably more effective at exploiting soil resources (Norby, 1994), suggesting an initial advantage for Carex plants. However, this did not lead to an increased RGR of this species.
SLA and LMF
Based on numerous CO2 enrichment studies which have investigated carbohydrate accumulation (Schäppi & Körner, 1996; Poorter et al., 1997) it is likely that the lower observed SLA in the high CO2 treatment was a result of accumulation of carbohydrates in the leaves. This was documented for Bromus and Carex at our field site as well (Obrist et al., 2001).
Our data do not meet Roumet & Roy’s (1996) hypothesis of a high SLA favouring a strong growth response to CO2 enrichment. Carex plants had both a lower SLA and a stronger biomass response than Bromus plants. Instead, higher LMF and higher leaf longevity of Carex may have compensated the seeming disadvantage (Poorter & van der Werf, 1998).
Photosynthesis and leaf senescence
The observed photosynthetic stimulation was similar to the +31% for Bromus and +43% for Carex obtained in the field (D. Spinnler, unpublished), suggesting that our experimental conditions closely matched the natural situation. Because CO2 stimulation is commonly lowest and downward adjustment of photosynthesis strongest on nitrogen deficient soils (Stitt & Krapp, 1999), we suspect low soil N-availability to explain the moderate photosynthetic stimulation observed in both species. Sage et al. (1997) found a 23% decline of Rubisco content in Bromus compared to no such Rubisco response in Carex following high CO2 exposure at our reference field site. This observation may contribute to the slightly smaller photosynthetic stimulation noted in Bromus.
The greater dead leaf fraction in Bromus may have resulted from both greater leaf production and shorter longevity and is a commonly observed effect (Miller et al., 1997; Navas et al., 1997; Cook et al., 1998). Significant differences of the allometric coefficient (k) for senesced leaves, confirm this to be a true CO2 effect as opposed to a biomass induced (ontogenetic) effect. A high CO2 induced earlier decline of photosynthesis in Bromus may have caused the lower CO2 stimulation of photosynthesis compared with Carex.
We conclude that the CO2 effect on biomass of Bromus and Carex plants under the exclusion of periodic drought (and perhaps other natural growth constraints) is not as different as field observation had suggested. Since growth analysis revealed no significant CO2 effect on dry matter allocation, we assume that the remaining, ‘true’ CO2 effect was largely photosynthesis-derived, as this was the only factor with a significant beneficial CO2 effect. We suspect the genotype-specific CO2 effect on growth to depend on other environmental factors (e.g. water limitation and nutrient status) and the ontogenetic status (age, size) of the tested genotypes. We strongly advise against deducing evolutionary effective genotype properties from spot measurements in time, because genotypic responses to CO2 may not be sustained. If existent, such genotypic effects would be quite small in the species tested here, but even minute genotype effects (below the resolution of our trial) could significantly influence the future gene-frequency of these grassland species. A number of the observations made during this controlled environment study did not match responses seen in the field and thus underline the importance of environmental covariables for plant CO2 responses.
The authors wish to thank Tanya Handa and Dieter Spinnler for valuable help and discussion. This study was supported through the Swiss National Science Foundation as part of the Priority Programme Environment.