Author for correspondence: J.-S. He Tel: +1 617 496 8924 Fax: +1 617 496 5223 Email: firstname.lastname@example.org
•This study was conducted to determine whether elevated CO 2 alters patterns of plant reproduction, and whether density affects population- and individual-level responses to elevated CO 2 .
•Phytolacca americana was grown in a glasshouse at three population densities under ambient and elevated CO 2 environments, and harvested at both vegetative and seed mature stages.
•CO2 did not affect the observed or estimated minimum size required for reproduction. At the population-level, elevated CO2 increased the total and above-ground biomass at both harvests. Density decreased both measurements at the second harvest. At the individual-level, elevated CO2 increased reproductive mass but decreased seed size, and the responses of reproductive allocation were density-dependent. Net photosynthesis at saturating light (Pmax) increased under elevated CO2, but decreased with density, with a CO2 × density interaction.
•hese results indicate that CO 2 advances timing of flowering by changing growth rate rather than modifying minimum size required for reproduction, while density modifies the responses of reproductive allocations to elevated CO 2 in P. americana .
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Under elevated atmospheric CO2 concentrations, the future biodiversity of plant communities will depend on both changes in growth rates and reproductive success (Morison & Lawlor, 1999; Körner, 2000; LaDeau & Clark, 2001). If there is interspecific variation in the fitness under CO2 enrichment, then advantages gained by some species over others could have dramatic consequences for future communities (Jackson et al., 1994; LaDeau & Clark, 2001). Even if specific changes in reproduction are too small to detect, they may accumulate to large net effects (Gifford et al., 1996).
Studies that have incorporated density–dependent interactions suggest that CO2-induced growth enhancements are generally lower when individuals are grown in the presence of neighboring plants (Ackerly & Bazzaz, 1995; Wayne & Bazzaz, 1995; Retuerto et al., 1996; Wayne et al., 1999). When size hierarchies were compared between stands grown in ambient and elevated CO2 environments, the variance in size hierarchies was reduced under elevated CO2 (Wayne & Bazzaz, 1997) because elevated CO2 can cause the suppressed understory individuals to be more productive in biomass (Wayne & Bazzaz, 1995; Kerstiens, 2001). This could lead to an alteration of the effective population size as more of the otherwise suppressed individuals produce seeds. Therefore, it is important to study CO2 effect on populations rather than single individuals.
Several studies have distinguished between the expected responses of determinate and indeterminate plants to elevated CO2 (Lawlor & Keys, 1993; Morison & Lawlor, 1999). Determinate plants are considered to offer only limited sink capacity for assimilates because their developmental regulation is strictly controlled and the number and size of organs are limited and fixed (Lawlor & Keys, 1993). Indeterminate plants, on the other hand, do not suffer from developmentally restricted capacity for growth or storage (Ho, 1988). For example, the biomass response to elevated CO2 is greater in the plants that have nonfoliar starch storage organs, such as perennials and root crops (Pooter, 1993; Miglietta et al., 1998). Thus, we expected that an increase in nonfoliar vegetative storage due to elevated CO2 would promote reproductive output, if a positive linear relationship between reproductive output and plant size exists.
The objective of this study was to determine the effects of elevated CO2 and density on the reproduction and biomass allocation in Phytolacca americana, an indeterminate perennial. In particular, we wanted to examine whether elevated CO2 alters patterns of plant reproduction, that is the minimum size for reproduction, reproductive allocation and total reproductive biomass, whether stand-level and individual-level responses to CO2 differ; and if population density modifies the effect of CO2.
Materials and Methods
Phytolacca americana L. (Phytolaccaceae), commonly known as pokeweed, is a polycarpic perennial herb common to much of the eastern United States, ranging from Quebec and Ontario south to northeastern Mexico ( Caulkins & Wyatt, 1990 ). It is often abundant in open disturbed habitats, particularly in forest edges and canopy gaps ( Sauer, 1952 ; Caulkins & Wyatt, 1990 ; Wilson & Shure, 1993 ). According to our investigation at Harvard University's Concord Field Station, MA, the first-year plants have an average height of 98 cm, with an average total biomass of 28 g, and the root depth of 15–25 cm, depending on soil conditions. The plant is a predominantly autogamous species ( Armesto et al., 1983 ). P. americana was selected for this study because it is an edge plant. As forest areas diminish in the future, edge habitats will become increasingly more common.
Experimental design and growth condition
Mature seeds of P. americana were collected from several individuals of a population in Lexington, MA, in October of 1999 and stored dry at 4°C for 6 months. On June 29, 2000, seeds were sown in 53 × 40 × 20 cm plastic tubs (Consolidated Plastics Company, Inc, Twinsburg, OH, USA) filled with Pro-Mix general-purpose growing medium (Premier Horticultural Company, Red Hill, PA, USA). This soil has a background nutrient level of 70–150 mg L−1 NO3− N. To each tub, 9 g Osmocote (Scotts-Sierra Horticultural Products Company, Marysville, OH, USA) controlled release fertilizer (N: P: K = 14%: 14%: 14%), which released evenly over a 4-month period, was applied when seeds were sown. No further fertilizer was added during the experiment. Six holes were drilled into the bottom of each tub to provide drainage. Three density levels and two CO2 levels (3 density × 2 CO2) were applied in a complete factorial design. Tubs were randomly assigned to one of the 6 treatment combinations. There were 18 replicates per treatment (n = 108). Two CO2 concentration levels (370 or 700 µmol CO2 mol−1) simulated atmospheric CO2 concentrations of the present and the future. For the density treatments, 20, 100, and 500 seeds per tub were sown for the low, medium and high density, and produced an average of 8, 23 and 108 plants per tub (44, 126 and 592 plants m−2) at the harvest (Table 1). These density levels were typical of seedlings as observed for P. americana growing in open habitats and forest edges. No thinning was carried out during the experiment.
Table 1. Seeds sown and plants produced in the three density treatments. The mean number, standard error (SE), minimum and maximum of plants produced are listed
Tubs were randomly assigned to an environmentally controlled (including CO2) glasshouse at Harvard University (Cambridge, MA), which is divided into six separately controlled chambers. In three chambers, CO2 concentration was maintained at 370 µmol CO2 mol−1, while CO2 was maintained at 700 µmol CO2 mol−1 in the other chambers. The temperature in all chambers was kept at 25°C from 08.00 to 20.00 hours and at 19°C overnight. Lighting was provided by natural sunlight filtered through the roof of the glasshouse, which reduced light levels by about 28%. Tubs were arranged on one bench with the sides adjacent to form a population in each chamber, resulting in a rectangle of 3 × 6 tubs. Within each chamber, positions of the tubs were re-randomized once a week to reduce variation in growing conditions in the first 4 wk. From the fifth week, when the canopy was closed, a shade cloth wall was set around the rectangles to eliminate edge effects. The height of the shade cloth wall was set level with the top of the canopy and adjusted as the canopy height increased. Plants were watered daily throughout the experiment.
The first flowering occurred in an elevated CO2 chamber on August 28. One week later, plants in ambient CO2 began to flower. From September 2–4, after some plants in both treatments had started to flower, one third of the plants from both CO2 treatments (n = 36) were harvested. To minimize edge effects, only 10 plants in the center of each tub of medium and high densities were used for further quantitative analysis. At low density, all plants in the tubs were used in order to maintain adequate and balanced sample size for quantitative analysis. However, the nontarget plants in medium and high density were used for the measurements of above-ground and below-ground biomass of the tubs. Leaf area was measured with a LI-3100 Area Meter (Li-Cor Inc., Lincoln, NE, USA). Roots were rinsed to remove soil particles. Plant material was dried to a constant weight at 65°C and weighed on an Acculab Lt-320 balance (Danvers, MA, USA). Dry weight was used to determine the biomass allocation. Specific leaf area (cm2 g−1) was determined by dividing leaf area by the leaf dry weight of each plant.
The remainder of the plants (n = 72) reached the seed mature stage between November 15–22, at which point they were harvested, dried and weighed as above. In addition, fruit weight was determined for each plant and each tub. The total number of seeds produced per plant was determined by multiplying the mean number of seeds produced per 1 g of random fruits on that plant by the dry mass of fruits on that plant. Mean seed mass was determined by dividing seed weight, measured on an electronic semianalytical balance (Sartorius AG, Goettingen, Germany), by the seed number of 1 g of random fruits on each plant. Dry weights were used to calculate the percentage of total biomass allocated to roots, stems, leaves and reproductive structures for each plant.
Net photosynthesis at saturating light (Pmax)
Pmax was measured at the full-flowering stage, which was developmentally determined for each treatment, using an open path gas-exchange system (Li-Cor 6400) with a red-blue light source and a CO2 mixer (Li-Cor Inc., Lincoln, NE, USA). For each treatment, we measured two tubs, and three individuals in each tub. During all measurements, temperature in the leaf cuvette was maintained at 25°C and relative humidity was kept between 50 and 65%. Reference CO2 concentrations were maintained during measurement at 370 µmol CO2 mol−1 in ambient CO2 and 700 µmol CO2 mol−1 in elevated CO2. The saturating photosynthetic photon flux density (PPFD) was 1500 µmol m−2 s−1.
The reproductive mass of P. americana showed a highly positive linear relationship with plant total biomass (R2 > 0.84, P < 0.0001) at all three densities. A simple model F=a * (B – b) (Weiner, 1988) was used to estimate the flowering size, that is the minimum size required for reproduction, where F is the reproductive mass at ambient or elevated CO2, B is the total biomass, a is slope of the relationship, and b is the flowering size. We used the method of Zar (1999) to compare simple linear regression equations under ambient and elevated CO2. At the same time, the observed range of flowering size (height and total biomass) at ambient or elevated CO2 was determined by connecting the size of the smallest flowering plant with the size of the largest non-flowering plant of the population (Wesselingh et al., 1997). The differences in the size distributions of fruit biomass and total biomass for the flowering individuals at ambient and elevated CO2 concentrations were tested using the Kolmogonov-Smimov nonparametric method (Sokal & Rohlf, 1995).
The data were analyzed separately at population (tub) and individual level using a two-factor multivariate analysis of variance (MANOVA) to test for the effects of CO2 and density on the variables for population biomass and individual reproductive characteristics, because we measured a number of dependent variables and were interested in looking at the overall response across the suite of variables (Meekins & McCarthy, 2000; Scheiner & Gurevitch, 2001). The Wilks’ Lambda was used to test for significance of each MANOVA. To determine which variable or variables was responsible for the difference in the CO2 and density treatments, those variables that were analyzed in the MANOVA were also analyzed by subsequent ANOVA, using General Linear Model procedure, employing type III sums of squares. Significant results were explored using Scheffépost hoc tests for subsequent multiple comparisons. The MANOVAs and ANOVAs were performed using SAS version 8.01 (SAS Institute, 1999). All biomass variables were log10-transformed, and percentage data were arcsine transformed to meet assumptions of normality and homogeneity of variances.
For the first harvest (vegetative), results from the MANOVA indicated that only CO2 had a significant effect (Table 2). Further ANOVA showed that CO2 mainly affected total and above-ground biomass (Table 3). Elevated CO2 significantly increased total biomass (Fig. 1a) and above-ground biomass (Fig. 1b). CO2 showed the trend of increasing below-ground biomass (Fig. 1c), but the effect is statistically insignificant. RSR was not affected by the two factors (Table 3, Fig. 1d). The CO2 × density interaction was insignificant.
Table 2. Results of multivariate analysis of variance (MANOVA) for the effects of CO 2 and density on total, above-ground, below-ground biomass, and root : shoot ratio (in both harvests), and fruit biomass and flowering ratio (Harvest 2) measured for populations of Phytolacca americana . The parameters df (H) and df (E) denote the degrees of freedom for the hypothesis and error sum of squares cross product matrices, respectively
Table 3. F values resulting from a two-factor analysis of variance (ANOVA) on total, above-ground, below-ground biomass, and root: shoot ratio (in both harvests), and fruit biomass and flowering ratio (Harvest 2) of Phytolacca americana populations grown under ambient and elevated CO 2 and at low, medium and high densities. Degrees of freedom in the model were as follows: CO 2 (1), density (2), and CO 2 × density (2). Significant symbols are as follows: * P < 0.05, ** P < 0.01
At the second harvest (mature), results from the MANOVA indicated that both CO2 and density had significant effects (Table 2). Further ANOVA showed that CO2 and population density significantly affected total biomass, above-ground biomass, fruit production and RSR. Below-ground biomass was not affected by the two factors (Table 3). Elevated CO2 significantly increased total biomass, above-ground biomass (Fig. 1e,f), and population fruit production (Fig. 2a). Medium and high density significantly increased RSR (Fig. 1h). There was no interaction between density and CO2.
Only density influenced population flowering ratios (proportion of flowering plants which fruited) (Table 3). Flowering ratio declined from 79.7% at low density to 21.9% at high density (Fig. 2b). Elevated CO2 had the trend to increase the flowering ratio (Fig. 2b), but the effect is statistically insignificant (Table 3).
The Kolmogonov-Smimov test indicated that the size distributions of flowering individuals between ambient and elevated CO2 were different (P < 0.05). Flowering plants under elevated CO2 had a larger proportion of smaller plants (total biomass under 10 g) than those of ambient CO2, indicating that elevated CO2 mainly increased the flowering ratio of smaller plants. Furthermore, there were larger size variations at elevated CO2 than at ambient CO2 (Fig. 3).
Fruit production was significantly (P < 0.001) and strongly (R2 > 0.84) correlated with total biomass of the plants at all three densities (Fig. 4a–c). Only plants from high density were used to calculate the flowering size because of their biomass range (from 2.15 g to 32.51 g). There was no difference in the estimated flowering sizes between plants grown at ambient (3.89 ± 0.403 g) and elevated CO2 (3.71 ± 0.426 g) (P > 0.05) (Table 4). The observed ranges of flowering sizes of populations at ambient and elevated CO2 were 2.15–7.62 g and 2.18–11.61 g, respectively. There was no difference in the observed minimum flowering size. We also found that plants under a height of 0.61 m at ambient CO2 and 0.59 m at elevated CO2 did not bear flowers.
Table 4. Parameter estimates for the relationship between fruit biomass and total biomass at ambient and elevated CO 2 . The model F = a * ( B – b ) was used, where F is the fruit mass at ambient or elevated CO 2 , B is the total biomass, a is the slope parameter for the relationship, and b is the minimum size required for reproduction. n , number of samples. Because the relationship may depend on density, only samples of high density were used
b (x-intercept) (± SE)
a (slope) (± SE)
3.89 ± 0.403
0.293 ± 0.020
3.71 ± 0.426
0.280 ± 0.017
Results from the MANOVA indicated that elevated CO2 and density significantly affected resource allocation and reproductive characteristics at the individual level. The two–way interaction was also significant (Table 5). Further ANOVA indicated that elevated CO2 significantly increased reproductive mass per unit leaf surface area (estimate of fecundity as a function of the potential for carbon gain of the vegetative structures, see Huxman et al., 1999), and reproductive mass per plant (Table 6, Fig. 5b,d). Elevated CO2 marginally increased seed number (P = 0.077) per plant. The effects of elevated CO2 on reproductive allocation (reproductive mass per unit vegetative mass and reproductive mass per unit leaf surface area), were highly density-dependent (Fig. 5a,b). For example, the increase in reproductive mass per unit leaf surface area was observed only at low population density. The increase became less pronounced at high density (Fig. 5b). Elevated CO2 can either decrease or increase the reproductive mass per unit vegetative mass in P. americana depending on population density. It decreased it at low density, but increased it at high density, while there was no effect at medium density, showing a significant CO2 × density interaction (Table 6, Fig. 5a).
Table 5. Results of multivariate analysis of variance (MANOVA) for the effects of CO 2 and density on the reproductive characteristics of Phytolacca americana . The parameters df (H) and df (E) denote the degrees of freedom for the hypothesis and error sum of squares cross product matrices, respectively
Table 6. F values resulting from a two-factor analysis of variance (ANOVA) on reproductive characteristics of Phytolacca americana grown under ambient and elevated CO 2 and at low, medium and high densities. Degrees of freedom in the model were as follows: CO 2 (1), density (2), and CO 2 × density (2). Significant symbols are as follows: * P < 0.05, ** P < 0.01
Density had large influences on reproductive allocation and fecundity (Table 6). High density significantly decreased reproductive mass per unit vegetative mass, reproductive mass per unit leaf surface area, seed number per plant and reproductive mass per plant (Fig. 5a–d). Elevated CO2 significantly decreased seed size (Table 6).
Net photosynthesis at saturating light (Pmax) and specific leaf area
Across all three density treatments, Pmax increased 28% at elevated CO2 (P < 0.001, ANOVA results not shown) (Fig. 6a). The effect of elevated CO2 on Pmax was also density-dependent, with increasing ratios 1.43, 1.19 and 1.16 at low, medium and high density, respectively, showing a significant CO2 × density interaction (P < 0.001).
Elevated CO2 significantly decreased specific leaf area (SLA) at all three density treatments (Fig. 6b). Across all density treatments, SLA decreased on average from 463 ± 30.7 cm2 g−1 (Mean ± 1 SE) at ambient CO2, to 270 ± 17.2 cm2 g−1 at elevated CO2, decreasing 42%. The density treatment did not affect SLA, but it significantly decreased total leaf area per plant. Elevated CO2 slightly decreased total leaf area per plant (average total leaf area per plant was 3094 ± 373 and 2694 ± 325 cm2 under ambient and elevated, respectively), but the effect was not statistically significant (P = 0.344). There was no CO2 × density interaction in SLA or total leaf area per plant.
Elevated CO2 did not modify plant flowering size
There is little reason to expect increasing CO2 to alter ontogenetic development. Temperature, interacting with environmental conditions such as photoperiod, is a key modifier of ontogenetic development rates (Morison & Lawlor, 1999). This is the case in some crops (Rawson, 1992), but in other species, larger effects of CO2 on ontogenetic development have been observed. Depending on the species, elevated CO2 can increase, decrease, or have no effect on flowering development. For example, elevated CO2 caused plants to flower earlier in Polygonum spp., Cassia spp. (Farnsworth & Bazzaz, 1995), Centaurea jacea, Betonica officinalis (Rusterholz & Erhardt, 1998), Datura stramonium (Garbutt & Bazzaz, 1984), Amaranthus retroflexus (Garbutt et al., 1990), Senecio vulgaris and Poa annua (Leishman et al., 1999). In contrast, it caused other plants to flower later, such as Setaria retroflexus (Garbutt et al., 1990). Reekie & Hicklenton (1994) reported that CO2 advanced flower opening in four long-day species, but delayed flowering in four short-day species.
In the present experiment, we found that plants of P. americana at elevated CO2 begin flowering 1 wk earlier than at ambient CO2. Earlier flowering can be achieved through two mechanisms: modification of the flowering size, or alteration of the relative growth rate. Rather than modifying flowering size at which plants switch from vegetative growth to reproductive growth, CO2 appears to affect phenology by changing growth rate in P. americana. In a previous study, it was found that elevated CO2 decreased the number of leaves at flowering in Guara brachycarpa and Oenothera laciniata (Reekie & Bazzaz, 1991). We also noticed that if the number of leaves was measured as the critical size for flowering, two factors needed to be taken into consideration: the differences in specific leaf area (SLA), and the leaf size. SLA decreased with elevated CO2 in this study, as well as in the others (Bazzaz, 1990; Huxman et al., 1999). When flowering size was measured by total biomass, it is obvious that elevated CO2 did not modify the flowering size of P. americana.
The density dependence of plant responses in reproductive allocation to elevated CO2
Plant density is one of the important determinants of plant growth and reproduction (Harper, 1977). In a densely occupied habitat, there may be a decrease in individual plant biomass due to competition for resources (Grace & Tilman, 1990). Studies showed that the survivorship, the proportion of plants flowering and fruiting, the number of seeds per individual, the total seed production per population, and the mean seed mass, all declined with increasing density (Harper, 1977; Bazzaz et al., 1992). At the population level, Abutilon theophrasti grown under higher density had lower total biomass (Casper & Cahill, 1998). In P. americana, we observed that at population level the interaction between CO2 and density was insignificant, but at individual level, the effects of elevated CO2 on reproductive allocation were density-dependent. For example, elevated CO2 decreased the reproductive mass per unit vegetative mass at low density, but increased it at high density. Furthermore, elevated CO2 mainly increased the flowering ratio of smaller plants. These results support the hypothesis that elevated CO2 can lead to an alteration of the effective population size and population viability, as those suppressed species become more competitive in biomass production.
Farnsworth & Bazzaz (1995 ) showed that vegetative characteristics alone are poor predictors of fitness and future population dynamics. In their study, early vegetative growth responses to elevated CO 2 were not a strong predictor of subsequent reproduction in nine herbaceous annual species. Leishman et al. (1999 ) arrived at the same conclusion. However, the vast majority of agricultural predictive models use growth parameters alone to estimate future levels of carbon sequestration ( Goudriaan et al., 1999 ). In the present study, although total biomass is a good predictor for the reproductive responses to elevated CO 2 , it was accompanied by production of smaller seeds, especially at low density. From an evolutionary perspective, the smaller seeds produced under elevated CO 2 may have a far-reaching impact on reproductive success or fitness, as experiments have shown that larger-seeded species may have advantages under hazards like drought, mineral nutrient deficiency, dense shade, clipping, and burial under litter and soil ( Leishman et al., 2000 ). The density-dependent effects of CO 2 enrichment on relative reproductive success and fitness can significantly influence the dynamics and microevolution of natural populations in the future environments ( Curtis et al., 1994 ; Bazzaz et al., 1995 ).
Only a few researchers have evaluated the quality of seeds produced under elevated CO2 conditions. For example, smaller seeds produced from elevated parental CO2 growth conditions lead to seedlings that produce smaller leaves that are delayed in development and have smaller roots (Huxman et al., 1999). However, Steinger et al. (2000) found that although elevated CO2 increased seed mass, parental CO2 growth conditions had no significant effect on seedling size. They argued that the advantage of increased seed mass at elevated CO2 may be offset by the reduced concentration of nitrogen. The McGinley & Charnov (1988) model suggested that the optimal seed size should be positively correlated with the ratio of the carbon and nitrogen pools available for investment to offspring, and that there should be a negative correlation between seed size and absolute seed nitrogen content. Because low nitrogen content in seeds has been shown a consistent trend across other studies (Parrish & Bazzaz, 1985; Huxman et al., 1999; Leishman et al., 1999; Steinger et al., 2000). More but smaller seeds in P. americana might be caused by the trade-off between seed number and seed size (Bazzaz et al., 2000).
Indeterminate plants and RSR
The partitioning of resources to plant organs is the outcome of many processes (Reynolds & Thornley, 1982; Bazzaz, 1997). Models of root : shoot partitioning have been proposed which consider carbon and nitrogen supply and utilization as the driving variables controlling allocation and growth (Reynolds & Thornley, 1982; Levin et al., 1989; Grace, 1997). Optimal partitioning models predict that plants respond to environmental variation by partitioning biomass among various organs or structures to optimize resource acquisition and maximize growth (Thornley, 1969; Hirose, 1987; Levin et al., 1989; Bernacchi et al., 2000). When plants are exposed to elevated CO2, increased carbon acquisition results in a shift in allocation toward roots until root activity is proportionally enhanced. RSR is also influenced by the development of reproductive structures, which represent competing sinks for carbohydrates (Ho, 1988; Johnson & Lincoln, 2000). It is expected that indeterminate plants grown under elevated CO2 will have a proportionately greater allocation of assimilate to roots than plants grown under ambient CO2, and that they will have a larger response (Johnson & Lincoln, 2000). Apparently, this is not the case for P. americana, in which elevated CO2 had no effect on RSR at vegetative stage, but decreased it at seed maturation. Also contrary to prediction, three cultivars of Raphanus sativus and the wild, R. raphanistrum, differing in root to shoot ratios, did not differ in total biomass at mature stage under two levels of CO2 (Jablonski, 1997). Other studies also found that responses of biomass allocation to elevated CO2 were not consistent with optimal partitioning predictions (Bernacchi et al., 2000). However, our study supports the hypothesis that allocation adjustments in response to CO2 should be less pronounced or absent when soil resources are nonlimiting (Rogers et al., 1996; Curtis & Wang, 1998).
Early successional species such as P. americana often occur in monospecific patches of varying density. It is therefore crucial to study CO2 effects on populations of such plants rather than on single individuals. In all experimental populations, high CO2 caused P. americana to flower earlier. Changes in plant growth rate appear to contribute more to the CO2 effect than modifications in the size at which plants switch from vegetative growth to reproduction. Our study demonstrates that the responses to elevated CO2 differ between the stand-level and the individual-level, with the latter showing a density-dependent response. It is unclear whether these density-dependent responses are species-specific. If so, the species-specific responses could significantly influence the dynamics of natural populations and community composition in future elevated CO2 environments.
Both authors wish to thank summer students M. Villar, T. K. Ngodup, E. Berdan, and P. Sima for help carrying out the experiment, and R. Stomberg for managing the glasshouse facilities. We thank B. Schmid, G. Bauer, S. Kaufman, K. Wolfe-Bellin and D. Flynn for their very constructive comments on the manuscript. This research was supported by Harvard Forest LTER and Andrew Mellon Foundation.