Natural variation in stress response induced by low CO₂ in Arabidopsis thaliana

2.1 Plant materials and growth conditions

Thirty-one A. thaliana accessions were used in this study (Table 1) to represent a wide range of geographically separated locations, elevations, and climates.

Arabidopsis seeds were surface-sterilized by soaking in 75% (v/v) ethanol for 10 min and rinsed 5–6 times with 95% ethanol, then sown on solid media containing half-strength Murashige and Skoog mineral salts, 1% (w/v) sucrose, and 0.8% (w/v) agar, pH 5.7. Plates with seeds were incubated in the dark at 4°C for 2 days to break dormancy prior to germination in growth chambers. The 7-day-old seedlings were transferred to a mixture of perlite/vermiculite/peat (1:1:3) in a 8 cm square pot (512 cm³). For each CO₂ condition, at least 50 seedlings for one accession were transferred into the pot. Plants were then grown in a Percival controlled environment (E-36L, USA) growth chamber either at low CO₂ (100 ppm) or ambient CO₂ (380 ppm) with a 16 h light (22°C)/8 h dark (18°C) photoperiod and a light intensity of 120 µmol m⁻² s⁻¹ and 70% humidity. The CO₂ concentration was set the same as our previous study [30]. Four chambers, two for low CO₂ and the other two for 380 ppm CO₂, were used. The plants in the two (under the same condition) chambers were switched twice a week.

2.2 Growth parameters

Boyes et al. [32] defined 30 growth stages, which were divided into 9 principal stages for Arabidopsis, spanning development from seed imbibition through the completion of flowering and seed maturation. Based on the physiological growth stages of A. thaliana established by Boyes et al. [32], we chose stage 5.10 (first flower buds visible) and stage 6.00 (first flower open) to measure the growth parameters. At the beginning of stage 5.10, the transition from vegetative growth to reproductive growth, we recorded the number of days since germination until the first flower buds were visible, as well as the aboveground fresh weight (FW), number of rosette leaves, SLW, and the δ¹³C value in leaves. At the beginning of stage 6.00, we again recorded the number of days between germination and the opening of the first flower, as well as SD. Individual leaves were detached from each plant with forceps and imaged for subsequent analysis using a scanner (V900; Shanghai MICROTEK Technology Co. Ltd, Shanghai, China). Length and area were measured using IMAGE J (v1.8.0, https://imagej.nih.gov/ij/index.html) software.

2.3 Stable carbon isotope analysis

The fully expanded third true leaf of each plant that developed before stage 5.10 was used to quantify the stable carbon isotope ratio (¹³C/¹²C). All samples were oven-dried at 65°C for 48 h to a constant weight. The measurements of stable carbon isotope ratios were carried out at the Chinese Academy of Forestry’s Stable Isotope Laboratory (Beijing, China) using a Flash EA1112 HT elemental analyzer (Thermo Fisher Scientific, Waltham, MA, USA) coupled with a Delta V advantage isotope ratio mass spectrometer (Thermo Fisher Scientific). Stable carbon isotope ratios were expressed as δ¹³C (‰) and were calculated as follows:

where R_sample and R_standard are the ratios of ¹³C/¹²C in the samples and the standard (Pee Dee Belemnite), respectively. The precision of the repeated sample was 0.15‰.

2.4 SD measurement

The largest, fully expanded leaves were selected for SD measurement and prepared as follows: (1) leaves were fixed overnight or longer in FAA solution (5 mL of formaldehyde:5 mL of acetic acid:90 mL of 70% ethanol); (2) leaves were decolorized in 70% ethanol until white; (3) tissue samples were mounted abaxially on slides with Hoyer’s solution; and (4) stomata were visualized by differential interference contrast microscopy on a Zeiss Imager Z2 microscope (Carl Zeiss Microscopy, LLC, White Plains, NY, USA) (0.379 mm² field of view). Ten images were collected from the middle of the abaxial side of each leaf sample, between the mid-vein and the edge. Stomata were manually counted for all pictures and all leaves using IMAGE J (v1.8.0, https://imagej.nih.gov/ij/index.html). Six leaves per accession were analyzed.

2.5 Statistical analysis

Statistical analyses for all experiments were performed using Excel 2010 (Los Angeles, CA, USA), SPSS 19.0 (SPSS Inc., Chicago, IL, USA), and SigmaPlot (SyStat Software, San Jose, CA, USA) software. After calculating averages, standard deviations and standard errors were also determined. Significant differences between low and ambient CO₂ treatments for each trait and the interaction effect of CO₂ and accessions were determined by one-way analysis of variance with p ≤ 0.05 for each experiment.

3.1 Effect of treatment on traits

In this experiment, the low CO₂ concentration was set to 100 ppm, which was shown to be a severe stress to Arabidopsis ecotype Columbia-0, and the ambient CO₂ was set to 380 ppm the same as those in our previous study [30]. A collection of 31 accessions (Table 1) was selected to analyze the genetic diversity based on the whole set of measurable responses to low CO₂. As expected, all tested accessions showed reduced growth when grown under low CO₂ versus ambient CO₂ (380 ppm) (Data not shown. Part results are shown in Figure A1).

As shown in Table 2, the effects of CO₂ concentration and accession were strongly significant in the comparison of the number of days to stage 5.10, FW, number of rosette leaves, SLW, and SD. The interaction effect of CO₂ and accession on these five traits was also highly significant.

3.2 Variation in flowering time

The onset of flowering, which is the transition from vegetative to reproductive stages, is a major determinant of a plant’s reproductive success and may be hastened or delayed by variations in climate that act as environmental cues or stimuli for the plant [33]. We measured the time from germination to the appearance of the first visible flower bud (developmental stage 5.10 as described in [32]) and the time to the first flower opening (developmental stage 6.00; [32]) of 31 accessions grown under low CO₂ and ambient CO₂, and calculated the difference in flowering times between the two CO₂ treatments. Two accessions, Mc-0 and Rsch-4, made the transition to flowering (stage 5.10) 4 days earlier under low CO₂ than under ambient CO₂ (Figure 1a and b). In 17 accessions, the low CO₂ treatment delayed flowering (stage 5.10) for at least 1 day. Among these, Ts-1 took 54 days longer to reach stage 5.10 under low CO₂ (Figure 1a and b). Te-0 and Kas-1 never flowered and died under low CO₂, so the data from these two accessions were missing in the following analysis. Twelve accessions (Figure 1b, red arrows) showed no difference in the time to stage 5.10 under low CO₂ and ambient CO₂, including Est, Ws-0, and Ler-0.

Figure 1

Effect of low CO₂ on flowering time. (a) Days from germination to stage 5.10. (b) The difference in time from germination to stage 5.10 or 6.00 for plants grown under low CO₂ compared with those under ambient CO₂. Values are mean ± SE (n = 3). Red arrows indicate the accessions with no difference in duration from germination to stage 5.10 between the two CO₂ treatments. Blue arrows indicate the accessions exhibiting a shorter time in days to reach stage 6.00 between the two CO₂ treatments. Arabidopsis accessions listed on the x-axis (left to right) are arranged by altitude, in the same order as in Table 1.

The time to the appearance of an open flower (stage 6.00) was far more variable than the time to stage 5.10, even though the timing of flower opening was consistently delayed in all the accessions when grown under low CO₂ (Figure 1b). This delay in the first flower opening ranged from 1 day (Rsch-4) to 63 days (Lip-0). Five accessions, including An-1, Pa-1, Mt-0, Ws-0, and Bay-0, showed less difference between the time to stage 6.00 under low CO₂ and ambient CO₂ (Figure 1b). Two accessions Est and Ga-0 died after reaching stage 6.00 in low CO₂ conditions. Under low CO₂, the first flower of Est opened partly but withered gradually and died, while a portion of the Ga-0 flower buds opened but had no seed in siliques and also subsequently died. The flower buds of Ts-1 failed to open under low CO₂ condition (Figure A1). Given the importance of a consistent flowering time when all conditions are stable except CO₂, we postulated that accessions that were able to maintain their time of flowering in spite of low CO₂ exhibited higher adaptability than accessions with a greater difference in flowering time. Supporting this point, five accessions failed to flower successfully and died at stages 5.10 and 6.00, indicating that they were unable to pass this developmental stage under low CO₂.

3.3 Aboveground biomass

Biomass is frequently used as a reliable estimate of plant fitness [34]. All the accessions tested in this study exhibited a reduction in plant size during low CO₂ growth. We measured aboveground biomass at the time of flowering (stage 5.10) and found that the aboveground (shoot) FW of all accessions decreased significantly (p < 0.001) under low CO₂ compared to biomass of plants grown under ambient CO₂ (Figure 2a). Two accessions, Pa-1 and An-1, showed a 60% reduction and four accessions showed a 70–80% reduction in shoot FW. The percent decrease for 7 accessions was between 80 and 90%, and for 16 accessions, biomass decreased over 90% (Figure 2a).

Figure 2

Effect of low CO₂ on shoot biomass. Relative FW, the ratio of FW under low CO₂ to FW under ambient CO₂. Red arrows indicate the accessions screened out by shoot biomass. Values are mean ± SE (n = 3). Stars denote significant differences between Col-0 and other accessions (*p < 0.05, ***p < 0.001).

We also calculated the variation in relative FWs between the two treatments by determining the ratio of shoot FW under low CO₂ to normal CO₂. We found that compared to Col-0, the accessions An-1, Est, Pa-1, Ws-0, Ler-0, and Sha all showed lower variation in relative FW when grown in CO₂-limiting conditions (Figure 2b). As with flowering time, we considered lower variation in FW for plants grown under low CO₂ compared to ambient CO₂ to be an indicator of higher adaptability by these accessions.

3.4 Rosette leaf number

The leaf number is closely correlated with the time to flowering and can be used as an indicator of phenotypic variability among different Arabidopsis accessions [35]. We counted the number of leaves in the rosette (excluding cotyledons) at the time of the first visible flower bud. Under low CO₂, most of the accessions bolted, resulting in fewer rosette leaves. For example, Be-0, Tsu-1, Mc-0, and Rsch-4 exhibited a greater than 50% reduction in leaves compared to those growth in ambient CO₂. However, An-1, Est, Pa-1, Ws-0, Ler-0, and Sha showed only a slight difference in rosette leaf number between treatments. Specifically, the leaf number of An-1 in low CO₂ was slightly greater than under ambient CO₂, whereas the other five accessions had on average one leaf less when grown under low CO₂ (Figure 3).

Figure 3

Effect of low CO₂ on rosette leaf number. Red arrows indicate the accessions with the least difference in leaf number between the two CO₂ treatments. Values are mean ± SE (n = 3).

We also counted the number of cauline leaves present at the time of the first flower opening. The cauline leaf response to low CO₂ was more variable among accessions than the rosette leaf response. On average, the number of cauline leaves was reduced under low CO₂ (data not shown), although in contrast, Wt-5 and Lip-0 had more cauline leaves due to the longer developmental time prior to reaching stage 6.00 from stage 5.10 under low CO₂. These two accessions also had more lateral branches.

Interestingly, there were four accessions for which the number of cauline leaves was less than 20% higher in low CO₂, whereas An-1 increased by 60% in low CO₂ compared to plants grown without CO₂ limitation. In contrast, 12 accessions exhibited a reduction in cauline leaves of less than 20% under low CO₂ and 9 accessions had 20–66% fewer leaves under CO₂-limiting treatment. The An-1, Est, Pa-1, Bay-0, Sha, and Wt-5 accessions had less than two leaves under ambient CO₂, resulting in percent difference of less than 20% except for An-1 and Wt-5. Since the role of cauline leaves in photosynthetic productivity is less certain than for rosette leaves given their typical variability under unmodified atmospheric CO₂, the contribution of variability in production of these leaves is also less predictable than that of rosette leaves, though in either case, we hypothesize that low variability indicates higher adaptability to low CO₂.

3.5 SD

Stomata are present on the leaf surface and control the entry of CO₂ into the leaves of plants prior to assimilation via photosynthesis [36,37,38]. Previous studies have reported that in Arabidopsis ecotype Col-0, SD (the number of stomata per unit leaf area) increased in response to low CO₂ concentration [30]. Here, we examined the SD of the abaxial (lower) leaf blade epidermis of the surviving Arabidopsis accessions grown under ambient and low CO₂ conditions. Among these 29 Arabidopsis accessions, there were 14 whose SD was significantly higher under low CO₂ compared to ambient CO₂ (Figure 4). Lc-0, Lm-2. Rsch-4, Ws-0, Bs-1, and Kil-0 did not show any significant differences in SD compared with Col-0. However, the SD of several accessions, including An-1 and Ler-0, decreased in response to low CO₂ (Figure 4). We are inclined to speculates, in light of these results, that accessions showing increased SD have higher fitness under low CO₂, since the higher number of stomata can increase the rate of CO₂ diffusion in leaves.

Figure 4

Effect of low CO₂ on SD. Relative SD is the ratio of the SD under low CO₂ to the SD under ambient CO₂. Red arrows indicate the accessions with no significant difference in relative SD compared to Col-0. Values are mean ± SE (n = 3). Stars denote significant differences between Col-0 and other accessions (*p < 0.05, ***p < 0.001).

3.6 SLW

SLW is defined as unit weight per unit leaf area, and it is an indicator of plant photosynthetic capacity, with high SLW interpreted as a decrease in photosynthetic efficiency [39,40,41,42]. In general, at low CO₂, SLW was lower than at ambient CO₂ among the Arabidopsis accessions in this screen. Here, we used the relative SLW, or the ratio of SLW under low CO₂/SLW under ambient CO₂, to evaluate the photosynthetic adaptations in response to low CO₂. Compared to Col-0, the accessions An-1, Est, Ws-0, and Ler-0 showed substantially lower variation between two treatments (Figure 5).

Figure 5

Effect of low CO₂ on SLW of accessions. Relative SLW is the ratio of the SLW under low CO₂ to the SLW under ambient CO₂. Red line delineates the relative SLW of Col-0. Red arrows indicate the accessions with higher relative SLW compared to Col-0. Values are mean ± SE (n = 3). Stars denote significant differences between Col-0 and the other accession (*p < 0.05, ***p < 0.001).

3.7 Stable carbon isotope ratio of leaf tissue (δ¹³C)

The stable carbon isotope ratio is used to distinguish the photosynthetic CO₂-fixing pathway in plants [43,44]. The δ¹³C values of C3 plants are typically more negative than those of C4 plants (−23 to −32% vs −6 to −19%, respectively) [43,44]. However, Arabidopsis carries some genes that belong to the C4 pathway, leading us to hypothesize that under ambient CO₂, this species may exhibit a less negative δ¹³C value under low CO₂ than ambient CO₂. To investigate the effect of low CO₂ on the photosynthetic capability of Arabidopsis accessions, we measured δ¹³C values in the third true leaves of all accessions, in order to analyze the stable carbon isotope ratio during treatment with low CO₂. Unexpectedly, the δ¹³C values of seven accessions at low CO₂ were more negative compared to those at ambient CO₂, whereas most of the other accessions had more positive δ¹³C values under low CO₂ treatment (Table 3), including the accessions An-1, Est, Ws-0, and Ler-0, thus suggesting a potential role for C4 genes in future potential adaptations to low CO₂.

3.8 Genetic background

From the aforementioned results, the four accessions An-1, Est, Ws-0, and Ler-0 showed less variation in flowering time, shoot biomass, rosette leaf number, and SLW between the two treatments than did Col-0 (Figures 1–3 and 5). Supporting these data, all four of these accessions exhibited smaller overall size compared to Col-0 under ambient CO₂ (Figure 6A). Our previous research [30] showed that low-CO₂ treatment upregulated some C₄-cycle genes including PEPC [45] and PEPC-K in Arabidopsis accession Col-0. The 1001 Genomes Project (https://1001genomes.org) provided whole genome sequence data to interrogate for genetic differences between different accessions, thus allowing us to potentially decipher how phenotypic variation is related to underlying genetic variation. We used the tool POLYMORPH (http://tools.1001genomes.org/polymorph/) to examine if low-CO₂-responsive genes carried specific sequence changes among the four accessions we screened with the most extreme responses to low CO₂. We calculated all polymorphic variants, including single nucleotide polymorphisms (SNPs), insertions, and deletions in all C₄-cycle genes and C₄-related transporter genes in the An-1, Est, Ws-0, and Ler-0 four accessions. However, no clear pattern in genetic variation emerged to indicate the mechanisms driving these phenotypic responses (Figure A2). For example, although PEPC (At2g42600) showed a 2.10-fold higher transcript abundance in Col-0 in response to low CO₂ [30], An-1, Est, and Ws-0 had no identifiable differences in PEPC sequence compared to that of Col-0 (Figure 6b). Responses to low CO₂ stress involve a complex network of regulatory elements to participate in mitigating damage induced by the stress, and differences in genetic background may potentially trigger unique stress responses among different accessions. Using transcriptomics sequence data, Carlson et al. [46] determined that a significant number of SNPs were absent in two accessions of Arabidopsis suecica (a relatively recent allopolyploid species) in the 1,001 genome SNP collection. RNA-seq analysis can effectively identify the genetic variation among these four accessions and we will use this technique in further experiments to explore the genetic basis underpinning plant adaptation to low CO₂.

Figure 6

Phenotype and genetic difference of the four accessions An-1, Est, Ws-0, and Ler-0. (a) Phenotype of An-1, Est, Ws-0, and Ler-0 grown in ambient CO₂ or low CO₂. Scale bar = 2 cm. (b) The number of polymorphic variants (deletions, insertions, and SNPs) of the C₄-cycle PEPC gene in accessions An-1, Est, Ws-0, and Ler-0 when compared to the Col-0 reference genome. The calculation was performed using the tool POLYMORPH (http://tools.1001genomes.org/polymorph/).

4.1 Arabidopsis showed substantial natural phenotypic variation among wild accessions in response to low CO₂

Flowering time is an important determinant of plant fitness and represents a discrete developmental transition in response to environmental conditions [26]. Shifts in flowering time in response to low CO₂ availability have previously been observed for some Arabidopsis lines [23,30]. In this study, we observed significant variation in flowering time across accessions. Compared to ambient conditions, 12 accessions, including An-1, Est, Ws-0, and Ler-0, took the same amount of days to reach stage 5.10 under low CO₂. Only two accessions, Mc-0 and Rsch-4, flowered earlier (by 4 days) under CO₂ limitation, whereas among the 17 late-flowering Arabidopsis accessions, Ts-1 took 54 days more to reach stage 5.10, and Te-0 and Kas-1 died under low CO₂ without ever flowering (Figure 1a and b). The delay in first flower opening thus ranged from 1 day (Rsch-4) to 63 days (Lip-0). Two other accessions, Est and Ga-0, died after stage 6.00 under low, but not ambient CO₂.

Rosette leaf number is commonly used as a standard indicator of flowering time in Arabidopsis and with late flowering plants typically developing more rosette leaves. In our screen, six accessions An-1, Est, Pa-1, Ws-0, Ler-0, and Sha showed a slight difference in leaf number between treatments, whereas some accessions (Tsu-1, Be-0, Rsch-4, and Mc-0) showed a greater than 50% reduction in rosette leaf number (Figure 3). There is a very strong correlation between the time to flowering and the number of leaves at flowering in Arabidopsis [35], with previous study suggesting that these two traits may be genetically linked in wild accessions [48]. In our experiment, low CO₂ treatment delayed flowering time but did not increase the rosette leaf number, although a logical expectation would be that leaf number continues to increase as the duration of vegetative growth prior to flowering is prolonged. This finding agrees with results reported by Salomé et al. [48] for an F2 population derived from natural accessions in which the traits of “days to flower” and “leaf number” were canalized in natural accessions, though the link between the two could be genetically uncoupled. Taken together, these results suggest that response to low CO₂ entails a combination of physiological and morphological changes that maximize the efficiency of carbon assimilation in order to maintain consistent reproductive processes.

Changes in CO₂ concentration can induce profound effects on plant growth because of the central necessity for CO₂ in plant metabolism. Higher atmospheric CO₂ concentrations often boost the growth and reproduction of C₃ annuals, whereas low CO₂ has the opposite effect and decreases plant growth [30,49]. Previous studies showed that low CO₂ availability was a limiting factor in plant growth, leading to reduced production of plant biomass [18,23,30,50,51,52,53,54]. However, a delay of first flower opening was common among the 31 accessions under low CO₂ stress, the biomass of all accessions in our study decreased. Though all plants in this study, regardless of accession, exhibited a dwarfed morphology and decreased biomass when grown under CO₂-starvation conditions, we observed extensive variation in aboveground biomass, which ranged from 58% to greater than 95% lower biomass compared to their growth at full CO₂ availability (Figure 2). Furthermore, the accessions that grew the fastest under full, ambient CO₂ also showed the greatest reduction in biomass at low CO₂, consistent with previous research [55].

In general, subjecting plants to growth at 100 ppm CO₂ induced significant changes to vegetative growth and reproductive development across a range of phenotypic traits (Table 2), including later flowering, dwarf stature, reduced biomass, and reduced rosette leaf number, which varied among the Arabidopsis accessions. Thus, individual accessions may have developed an adaptive response to low CO₂ that can be used to further determine the genetic variability responsible for this adaptation.

4.2 Altitude of origin did not relate to low-CO₂ response

The partial pressure of atmospheric CO₂ decreases with the increase in altitude. Arabidopsis accessions adapted to growth at higher altitudes have presumably undergone a stronger selection for growth in lower CO₂ concentration than that of low altitude accessions. We hypothesized that adaptation to low CO₂ increases along an altitudinal gradient. To test this hypothesis, we used 31 Arabidopsis accessions that were originally collected from a variety of altitudes ranging from 50 to 3,400 m above sea level (Table 1). In Figures 1–5, the Arabidopsis accessions listed on the x-axis (left to right) are arranged by altitude, in the same order as in Table 1. However, we found that the responses to low CO₂ for all of the changes in measured traits in this study did not correlate linearly with altitude. For example, although all accessions had significantly lower aboveground biomass at low compared with ambient CO₂, six accessions An-1 (50 m), Est (50 m), Pa-1 (50 m), Ws-0 (150 m), Ler-0 (600 m), and Sha (3,400 m) performed much better than Col-0 (50 m) (Figure 2), suggesting that there was no clear differentiation between low-altitude genotypes and high-altitude genotypes. Ward and Strain [23] examined the responses to 20 Pa (200 ppm) CO₂ in eight accessions from seven different altitudes between sea level and 3,400 m and found that accessions exhibited limited heritable variation in the response of biomass production. Therefore, in this work, the altitude of origin did not significantly affect vegetative growth in response to low CO₂ (100 ppm).

4.3 Ws-0 was least affected by low CO₂

In this study, we found that the accessions An-1, Est, Ws-0, and Ler-0 showed less variation in flowering time, shoot biomass, rosette leaf number, and SLW between the two treatments compared to Col-0 (Figures 1–3 and 5). Compared to other accessions, their flowering time and rosette leaf number remained almost the same in the two CO₂ treatments, and shoot biomass was significantly less affected than in other accessions. Furthermore, the SLW was less affected by low CO₂, compared to accession Col-0. In light of the combined quantitative trait data, Est and Ws-0 were the least affected among these four accessions. As mentioned above, Est did not set seeds under low CO₂ (Figure A1), therefore Ws-0 would be the most effective candidate for further quantitative genetics studies.

When we examined the phenotypes of these four accessions grown under low CO₂, we found that they did not have a bigger plant size, whereas under ambient CO₂, they showed a smaller stature compared to other accessions. Temme et al. [55] reported species with fast growth or largest biomass at ambient CO₂ showed the strongest absolute reduction at low CO₂. Nitrogen content and photosynthetic rate are strongly affected by low CO₂ [18,56,57]. Previous studies have proposed that stress-tolerant plants have lower growth rates (reviewed in ref. [58]). One explanation of our observations is that their smaller stature and relatively slow growth under ambient CO₂ is an advantage in response to low CO₂. The small stature or slow growth among some plants may indicate low energy and low carbon demands, thus C–N cycle and photosynthesis may be less affected and these plants show less affect when grown under low CO₂. If this hypothesis is correct, it can provide us with valuable insight into the mechanisms by which C₄ metabolism arose and the reasons why it evolved independently in grasses (i.e., roughly half of the known C4 species are grasses) and also in a number of eudicots, for example, Amaranthaceae, Euphorbiaceae, Asteraceae, and Boraginaceae [59].

In this study, we compared the phenotypic variation in response to low (100 ppm) CO₂ among 31 Arabidopsis accessions to assess their relative adaptability through sustained vegetative growth and reproductive development. We found that A. thaliana displays extensive variation in its ability to adapt to low CO₂ and that this variation was correlated with their rate of growth under non-CO₂-limited conditions, rather than the altitude of origin for individual accessions. In particular, accession Ws-0 showed the least variability between treatments, indicating that it was the best potential candidate for use in further quantitative genetics studies and for isolation of genes underlying low CO₂ response. We also propose that a lower growth rate can attenuate the effects of low-CO₂ stress, though further experimental evidence is needed to test this hypothesis. Our findings on the physiological effects of growth under low CO₂ provide insight into the mechanisms by which individual plants and whole ecosystems may adapt to changes in atmospheric CO₂. As atmospheric levels of CO₂ rise, our increased understanding of these mechanisms governing carbon assimilation and flowering time during stress can improve our capability to predict the future of natural ecosystems subject to increasingly wide variations in climate.