SEARCH

SEARCH BY CITATION

Keywords:

  • CO 2 ;
  • glacial;
  • growth;
  • meta-analysis;
  • photosynthesis;
  • plant traits;
  • subambient CO2

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Conflict of Interest
  10. References
  11. Supporting Information

A general understanding of the links between atmospheric CO2 concentration and the functioning of the terrestrial biosphere requires not only an understanding of plant trait responses to the ongoing transition to higher CO2 but also the legacy effects of past low CO2. An interesting question is whether the transition from current to higher CO2 can be thought of as a continuation of the past trajectory of low to current CO2 levels. Determining this trajectory requires quantifying the effect sizes of plant response to low CO2. We performed a meta-analysis of low CO2 growth experiments on 34 studies with 54 species. We quantified how plant traits vary at reduced CO2 levels and whether C3 versus C4 and woody versus herbaceous plant species respond differently. At low CO2, plant functioning changed drastically: on average across all species, a 50% reduction in current atmospheric CO2 reduced net photosynthesis by 38%; increased stomatal conductance by 60% and decreased intrinsic water use efficiency by 48%. Total plant dry biomass decreased by 47%, while specific leaf area increased by 17%. Plant types responded similarly: the only significant differences being no increase in SLA for C4 species and a 16% smaller decrease in biomass for woody C3 species at glacial CO2. Quantitative comparison of low CO2 effect sizes to those from high CO2 studies showed that the magnitude of response of stomatal conductance, water use efficiency and SLA to increased CO2 can be thought of as continued shifts along the same line. However, net photosynthesis and dry weight responses to low CO2 were greater in magnitude than to high CO2. Understanding the causes for this discrepancy can lead to a general understanding of the links between atmospheric CO2 and plant responses with relevance for both the past and the future.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Conflict of Interest
  10. References
  11. Supporting Information

Atmospheric CO2 concentration has varied tremendously over geological time, from as high as 3000 ppm in the lower Devonian (Royer 2006) to as low as 180–280 ppm during the past 2.1 Ma of the Pleistocene (Honisch et al. 2009). About 17.5 Ka ago, atmospheric CO2 concentration started to rise from 180 ppm, levelled off at 280 ppm around 15 Ka ago and broadly remained at 280 ppm until the Industrial Revolution. Since the start of the Industrial Revolution, CO2 levels have risen to 390 ppm today, levels not experienced by plants for over 25 Ma (Royer 2006) and are expected to increase even further; common model estimates go up to 700 ppm by 2100 (IPCC 2007). A CO2 atmosphere of 700 ppm has not been observed since 42 million years ago (Royer 2006). The atmosphere today and as predicted for the end of the century is thus increasingly different from that experienced by plants during a large part of the recent past.

CO2 plays a pivotal role in a number of important ecophysiological processes: it is an essential ingredient for photosynthesis and plant growth, and it is highly likely that plants' morphological and physiological traits and their plastic responses to the CO2 concentrations are more tuned to the range of CO2 concentrations they have experienced recently. Because adjustment to changing CO2 involves changes in photosynthetic rates, nitrogen allocation, and other physiological properties (Curtis and Wang 1998; Korner 2000; Cowling 2001; Poorter and Navas 2003; Ainsworth and Long 2005), this trait adjustment has the potential to create a feedback that could affect the global carbon cycle (Beerling et al. 2012). Connecting the performance of plants of different species at low, ambient and future high CO2 is thus an important part of understanding the links between the atmospheric CO2 concentrations and the terrestrial biosphere in the past, present, and future.

At low CO2, photosynthesis is limited by the amount of available carbon by limiting carboxylase activity of the enzyme RuBisCO. Conversely, at higher CO2 concentrations, the rate at which CO2 can be taken up (photosynthetic capacity) becomes limiting (Sage 1994). CO2 levels in the past could thus have selected for RuBisCO carboxylase activity or efficiency. At current and future CO2 concentrations, other factors such as water and nutrient uptake will become more limiting than CO2 (Ward et al. 1999; Campbell and Sage 2006; Reich et al. 2006). This implies that trait states useful in a low CO2 atmosphere can be redundant or suboptimal in a high CO2 atmosphere. For example, high investment in RuBisCO, useful at low CO2, in a high CO2 environment requires a high N investment, which could otherwise be used in other N-limited steps. High activity of enzymes facilitating transport and binding of CO2 is a lower priority at high CO2 compared with the need for more sink capacity of photosynthates to take advantage of increased photosynthetic rates (Sage and Coleman 2001). Favorable traits in low CO2 thus do not necessarily mirror those in high CO2.

For obvious reasons, considerable scientific effort has gone into examining the response of plants to high levels of CO2 as projected for the latter half of this century. Several recent meta-analyses have found that, despite methodological differences among studies, a few main results are apparent: at high (500+ ppm) CO2, there is an increase in carbon assimilation and growth and decrease in stomatal conductance, nitrogen content and specific leaf area (Poorter 1993; Curtis and Wang 1998; Poorter and Navas 2003). The increase in biomass is about +45% for C3 species and +12% for C4 species at a 50% increase in CO2 concentration (Poorter and Navas 2003). The response of C4 species to increased CO2 is smaller than that of C3 species, probably because the carbon concentrating mechanism of C4 plants already concentrates CO2 around RuBisCO leaving less room for increased photosynthetic rate (Bowes 1993; Ghannoum et al. 2000). Overall, woody species showed a greater response to elevated CO2 than herbaceous species (Curtis and Wang 1998; Poorter and Navas 2003; Ainsworth and Long 2005; Lee et al. 2011).

In contrast to the large amount of studies on plant responses to elevated CO2, less research has been carried out on the response of plants to subambient, Pleistocene levels of CO2. Several individual experiments reveal that the influence of low CO2 acts on multiple biotic levels, ranging from leaf level to plant level and ecosystem level (Gerhart and Ward 2010). The emergence of agriculture has even been linked to the increase in CO2 to 280 ppm 17.5 Ka ago, as higher levels of CO2 lead to higher yields (Sage 1995). Understanding how plants have adapted to the low CO2 of their recent evolutionary history can aid us in understanding plants response to future high CO2 (Gerhart and Ward 2010; Beerling 2012; Leakey and Lau 2012). Recent research has shown CO2 uptake and water use are highly consistent across CO2, from low to high (Franks et al. 2013) Thus, there is clearly a need to integrate the knowledge available so far on low CO2 responses to determine whether more traits follow a predictable pattern.

Some qualitative expectations can be made as to how plants are likely to respond to low CO2. A lowering of CO2 will likely lead to a reduction in photosynthetic rates (Farquhar et al. 1980) and plant biomass (Overdieck et al. 1988; Cunniff et al. 2008). Next, an increase in leaf nitrogen concentration as RuBisCO may ameliorate some of the reduction in total C assimilation rate (Sage and Coleman 2001). Differences in response among different plant types can also be expected. C4 metabolism, which concentrates CO2 around RuBisCO, could partly compensate the potential reduction in growth as experienced by C3 plants. At lower atmospheric CO2 concentration, one of the mechanisms to maintain a high enough internal CO2 concentration is to open stomata wider, allowing water to escape at a faster rate (Farquhar and Sharkey 1982). Because of the carbon concentrating mechanism in C4 species, the diffusion gradient of CO2 across the stomata can be much steeper. At reduced CO2, this should allow C4 plants to maintain a smaller stomatal aperture than C3 plants, giving a smaller increase in stomatal conductance and a smaller decrease in water use efficiency (Farquhar and Sharkey 1982). As woody plants invest more biomass in stems than herbaceous plants (Poorter et al. 2012) and as stems usually do not contribute substantially to photosynthesis, it is possible that this constrains the ability to adjust carbon allocation at low atmospheric CO2. In response to a reduction in growth, a complex suite of trait adjustments, with differences among plant types, is expected at all physiological levels, varying from photosynthesis to biomass allocation. While the directions of all these responses to low CO2 have empirical support (Gerhart and Ward 2010), they have not yet been quantified in general terms across studies and species.

Here, we present the results of a global meta-analysis synthesizing data from currently available low CO2 experiments to quantify general patterns of morphological and ecophysiological trait responses to subambient CO2. In particular, together with the body of work on high CO2, we aim to build toward a general, quantitative understanding of the response of plant traits to a range of CO2.

Thus, we address the following research questions and hypotheses:

  1. How much do plant traits vary with decreased CO2 concentration?
    • We hypothesize that lower photosynthetic rates will lead to reduced growth at low CO2. To acclimate to a low CO2 environment and keep up photosynthetic rates, plants will have higher leaf nitrogen and larger stomatal conductance.
  2. How much do plant functional groups differ in their response to low CO2?
    • Because of the carbon concentrating mechanism of C4 plants, we hypothesize (a) that the negative effects of low CO2 on their photosynthesis and growth will be reduced as compared to those in C3 plants; and (b) that woody species will invest more of their biomass in nonphotosynthetic tissue leading to a greater reduction in biomass accumulation than herbaceous species.
  3. Is plant trait response to low CO2 similar in magnitude to the response to elevated CO2?
    • Atmospheric CO2 is on a trajectory from low during glacial times to very high CO2 in the future. We aim to shed light on whether plant traits adjust similarly from low to ambient as from ambient to high. Given the saturating nature of the photosynthetic response to CO2, we expect photosynthetic traits to respond to low CO2 through a greater magnitude shift compared with the high CO2 response. For other traits, we are curious if they follow the saturating response of photosynthesis or if they respond more proportional to CO2 changes.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Conflict of Interest
  10. References
  11. Supporting Information

We performed a literature review on plant science journals searching Web of Science using keywords “subambient CO2,” “low CO2,” “reduced CO2” and “glacial CO2.” This resulted in 33 papers that reported on studies with experimentally lowered atmospheric CO2 concentration for 54 species in total. In these experiments, plants were grown in greenhouses, climate chambers, or outdoor tubes after Mayeux et al. (1993). For the climate chambers and greenhouses, CO2 concentrations were reduced by passing air through some kind of filter or adsorbent (e.g., Soda lime) or through a reactive solution (e.g., NaOH).

For each study, we recorded sample size, duration, growing conditions, low CO2 treatment type, and germination conditions along with measures of plant physiological traits at the CO2 concentrations used (Table S1, Data S1 and S2). A full list of the studies found with which species and trait data they reported can be found in Table S1, Data S1 and S2. When data and errors were not present in tabular form, they were extracted from graphs using Datathief 3 (Tummers 2006). All papers reported trait means in response to CO2 concentration and most reported a measure of error (standard deviation, standard error, or confidence interval) for the trait in each CO2 treatment. Three of the published papers dealt with response to low CO2 at varying resource conditions (P limitation: Campbell and Sage 2006; Lewis et al. 2010; Drought: Ward et al. 1999). In order to exclude confounding factors, only those results at high nutrients and well-watered conditions were included. C3–C4 intermediates were grouped together with C4 species.

From the studies found, only 6 traits emerged with 10 or more species analyzed and only 20 traits with 3 or more species. Of these 20 traits, 12 were related to growth and development. These traits included specific leaf area (as SLA or leaf mass per area, which was recalculated to SLA, m2·g−1) and (components of) plant biomass (DW, g dry weight). In some cases, plant biomass was divided into above- and belowground mass. Aboveground mass was divided into leaf and stem mass. All of these masses can be expressed either in absolute terms or as allocation, that is, relative to plant mass. Number of stomata and stomatal pore size (μm) were infrequently reported. Five traits related to photosynthesis included photosynthetic rate either as maximum, at saturating light levels (Amax, μmol·m−2·s−1), and/or net, at growth conditions, photosynthesis (Anet, μmol·m−2·s−1), stomatal conductance (gs, mol·m−2·s−1), the ratio of internal to external CO2 concentration (Ci/Ca) and water use efficiency (WUE; Anet/gs, mmol·mol−1). Lastly, four traits were related to chemical composition, namely nitrogen content either area-based (g·m−2) or mass-based (g·g−1), chlorophyll content (μmol·g−1) and RuBisCO content (g·m−2).

To examine the effect of CO2 among all species in the study, we performed a weighted ANCOVA for each trait with CO2 concentration as covariate, species as a factor, and a potential interaction between the two. To determine the overall effect of CO2 on a plant trait, a model without differing slopes between species was fitted when species by CO2 interaction was not significant or the average slope from all species was calculated when species by CO2 interactions were significant. We used the inverse of the square root of the standard error from the original study as the weighting factor for uncertainty, as is standard in meta-analyses (Hedges and Olkin 1985; Sokal and Rohlf 1995) For trait values reported without an error term, the average standard deviation in the trait was used to calculate the weighting factor. All traits and CO2 concentrations were log-transformed, which improved the normality of the residuals and allowed the output to be considered as scaling slopes (Renton and Poorter 2011). We investigated both which traits responded to low CO2 and, for those traits that did, what the effect size of that adjustment was.

In the ANCOVA framework, the scaling slope of the trait-CO2 relationship then indicates the proportional change in trait value in the following way:

  • display math(1)

where CO2 change is the proportional change in CO2 concentration and β is the slope in the log–log plot. For example, if β were 1 then a 50% reduction in CO2 concentration would result in a 50% reduction in trait amount. When β is less or >1, a 50% reduction in CO2 will result in a less or more than 50% reduction in trait amount, respectively. A negative slope indicates an increase in trait value with a decrease in CO2 concentration. Subsequent to the ANCOVA analysis, differences in slope between C3 and C4 herbs and woody and herbaceous C3 plants were assessed by 2 sample t-tests weighted by 1/SE of the species. All statistics were performed using R, version 2.14.0 (R Foundation for Statistical Computing, Vienna, Austria). Due to the limited number of species for many traits, statistical power tended to be low; however, we judge it important enough to report those results as they reflect the current state of knowledge and to show the lack of data in important traits.

Rather than performing a meta-analysis on the available high CO2 experiments ourselves, we searched the literature for highly cited large-scale meta-analyses on plant traits in experimentally elevated CO2. From our survey, 5 large meta-analyses emerged involving tens to hundreds of plant species reporting various traits including the 6 traits that were reported for 10 or more species at reduced CO2 (Curtis and Wang 1998; Poorter and Navas 2003; Ainsworth and Long 2005; Ainsworth and Rogers 2007; Wang et al. 2012). From the meta-analyses, we extracted the shift in trait value at nonlimiting resources when available. We then compared this to the projected trait shift when assuming the same proportional response as to low CO2.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Conflict of Interest
  10. References
  11. Supporting Information

Of the 21 traits that were reported for 3 or more species, 14 showed either significant variation with CO2 or species response to CO2. Percentage values in the text below show the proportional change in trait value ± SE upon a 50% reduction in growth CO2 concentration (eqn 1). For each trait, species could respond to CO2 (adjust their trait value), show consistent variation in trait value between species over a CO2 gradient (species intercept or elevation of species line in trait vs. CO2 plot) and show significant variation in how species responded to CO2 (CO2*species interaction; Table 1). A nonsignificant interaction of species*CO2 for a given trait indicates that different species adjust the trait by the same proportional amount.

Table 1. Overview of ANCOVA results on log(trait data) versus log(CO2) concentration with species as covariate and as weighting factor. Traits are ordered by number of species analyzed. Slope indicates the average slope of log(trait) versus log(CO2) including SE. −50% CO2 gives the proportional change in trait given a 50% reduction in CO2 concentration as per Trait change = CO2 changeβ-1 where β is the slope. Values are calculated by slope ± SE
Trait#Species#StudiesSlope−50% CO2p(CO2)p(species)p(CO2*species)
  1. P-values are ns: not significant; †<0.1, *<0.05, **<0.01, ***<0.001.

WUE (mmol mol−1)2680.95 ± (0.1)−48.3% ± (3.5)*********
Anet (μmol·m−2·s−1)25150.7 ± (0.11)−38.3% ± (4.5)*********
DW (g)25140.91 ± (0.16)−46.9% ± (5.8)******ns
SLA (m2·g−1)2217−0.2 ± (0.08)+17.2% ± (6.4)******ns
gs (mol·m−2·s−1)1711−0.7 ± (0.13)+59.8% ± (13.9)*******
Amax (μmol·m−2·s−1)1590.58 ± (0.09)−33.1% ± (4.3)****ns
% Leaf N (g·g−1)109−0.2 ± (0.1)+17.8% ± (8)******ns
PNUE (μmol·mmol·N−1·s−1)1020.22 ± (0.36)−14.2% ± (21.6)*ns
r/s ratio930.34 ± (0.11)−21% ± (6.1)********
Shoot DW (g)740.62 ± (0.14)−35.1% ± (6.5)******ns
% Leaf mass54−0.1 ± (0.2) +9% ± (15)*******
Root DW (g)541.34 ± (0.27)−60.6% ± (7.5)******ns
Leaf DW (g)540.83 ± (0.54)−43.9% ± (21.5)********
C i /C a 44−0.1 ± (0.12)+6% ± (8.8)nsnsns
Chlorophyll (μmol·g−1)440.19 ± (0.09)−12.4% ± (5.4)ns***ns
Leaf N (g·m−2)440.23 ± (0.11)−14.8% ± (6.3)*nsns
% Stem mass430.08 ± (0.08)−5.4% ± (5)ns*ns
Stem DW (g)431.72 ± (0.83)−69.7% ± (18.5)*********
# Stomata420.08 ± (0.11)−5.7% ± (7.3)ns***ns
RuBisCO (g·m−2)330.37 ± (0.14)−22.7% ± (7.3)ns**ns
Pore size (μm)310.12 ± (0.13)−7.8% ± (8.2)ns***ns

Photosynthesis-related traits

Across the species studied, a 50% reduction in CO2 did on average reduce maximum photosynthesis (Amax, μmol·m−2·s−1) by 33 ± 4% (P < 0.05, 15 species; Fig. 1A) and net photosynthesis (Anet, μmol·m−2·s−1) comparably by 38 ± 5% (P < 0.001, 25 species; Fig 1B). Next to this, stomatal conductance (gs, mol·m−2·s−1) increased by 60 ± 14% (P < 0.01, 17 species; Fig. 1C). The ratio of water loss to carbon gain, intrinsic water use efficiency (Anet over gs, WUE), decreased by 48 ± 4% (P < 0.001, 26 species; Fig. 1D). For Anet, WUE and gs species showed significant variation in trait elevation (P < 0.001) and response to CO2 (interaction, P < 0.01). Amax however showed only significant variation in species trait elevation (P < 0.001).

image

Figure 1. Plant trait versus growth CO2 concentration (note the double-log scale) of the 6 traits with the highest number of species. (A) Maximum photosynthesis (Amax), (B) Net photosynthesis (Anet), (C) Stomatal conductance (gs), (D) Intrinsic water use efficiency (net photosynthesis over gs, WUE), (E) Plant dry weight, (F) Specific leaf area (SLA). Each line represents the response of a single species. Open symbols: C3 metabolism, solid symbols: C4 metabolism. Error bars give SE. Different colors represent the different families the species belong to.

Download figure to PowerPoint

Growth and allocation

A halving of growth CO2 concentration resulted in a corresponding reduction in plant dry weight (DW) by 47 ± 6% (< 0.001, 25 species; Fig. 1E). Aboveground biomass was less reduced than belowground biomass as Shoot DW was reduced by 35 ± 7% (P < 0.01, 7 species) and Root DW by 61 ± 8% (P < 0.001, 5 species). This pattern was reflected in a reduced root/shoot dry matter ratio (r/s ratio) of 21 ± 6% (< 0.001, 9 species). In contrast, specific leaf area (SLA, m2·g−1) increased by 17 ± 6% (< 0.001, 22 species; Fig. 1f) at reduced CO2. For root/shoot ratio, there was both significant variation in species trait elevation (P < 0.001) and response to CO2 (P < 0.01). For SLA and plant biomass, species showed significant variation in trait elevation but did not respond differently to reduced CO2.

Chemical composition

Only few data on chemical composition were available, strongly limiting statistical power of our test. The concentration of chlorophyll (μmol·g−1, 4 species) and amount of RuBisCO (g·m−2, 3 species) were not significantly affected by CO2. Nitrogen levels in the leaf showed a contrasting response to halving CO2 where leaf nitrogen percent (g·g−1) increased by 18 ± 8 (P < 0.001, 10 species) whereas nitrogen content per area (g·m−2) decreased by 15 ± 6% (P < 0.05, 4 species). Neither for nitrogen per unit mass nor per unit leaf area did species have significantly different responses to reduced CO2.

Differential responses of plant types

Due to the small number of available species, the statistical power of the comparison between plant types was limited. Thus, in a few cases, there was only a trend of differential response among C3, C4 and woody and herbaceous species. Nevertheless, interesting contrasts and similarities emerged. Figure 2 shows the contrasting slopes between plant types for the eight plant traits with nine or more species. Between C3 and C4 herbs, only the greater increase in SLA for C3 herbs at reduced CO2 was significant (P < 0.01). C4 plants showed on average a negligible SLA response to CO2. Net photosynthesis and dry weight seemed to be less reduced for C4 herbs although with small sample size and large variation this was not significant. Interestingly, plant dry weight of woody species was reduced less by 16% than that of herbaceous C3 plants (P < 0.05). Overall, for the traits shown in Figure 2, different plant types appear to show rather similar responses to reduced CO2.

image

Figure 2. Slope of log(trait) versus log(CO2 concentration) for different plant types, C3/C4 and woody/herbaceous. Open circles, C3 herbaceous type; solid circles, C4 herbaceous type; open square, C3 woody type. ±1 indicates a 1:1 change in a trait for a change in CO2 concentration with a negative slope indicating an increase in trait value and a positive slope indicating a decrease in trait value. Amax, maximum photosynthesis (μmol·m−2·s−1); Anet, net photosynthesis (μmol·m−2·s−1); gs, stomatal conductance (mol·m−2·s−1); WUE, water use efficiency (mmol·mol−1); DW, plant dry weight (g); SLA, specific leaf area (m2·g−1); r/s ratio, root DW to shoot DW (g·g−1). Numbers between brackets gives the number of species for each plant type. P-values between plant types are *< 0.05, **< 0.01.

Download figure to PowerPoint

Comparison with elevated CO2 experiments

Table 2 highlights the comparison between trait responses at low CO2 to the response at elevated CO2 of the six traits that had the most data. When comparing the trait shifts found in five large meta-analyses (Curtis and Wang 1998; Poorter and Navas 2003; Ainsworth and Long 2005; Ainsworth and Rogers 2007; Wang et al. 2012) to trait shifts extrapolated from the response to low CO2, a few interesting contrasts and similarities emerged. With increasing CO2, the magnitude of shift in Anet deviated more from the response to low CO2; for gs, the shift was similar in magnitude to results found in FACE studies and one growth chamber meta-analysis but not another. While we found no differences between C3 and C4 plants in water use efficiency at low CO2, at high CO2 large differences are found. The increase in WUE at high CO2 found for C3 plants at FACE sites is comparable to our extrapolated response. Whole plant dry weight (DW) appears to increase much less at high CO2 then expected based on the low CO2 response. SLA, however, seems to be adjusted in a similar magnitude as expected from the response to low CO2.

Table 2. Comparison of trait shift at high CO2 extrapolated from low CO2 response (bold values) to actual changes found in three meta-analyses. If the trait adjustments are proportional from past low to future high CO2, the predictions from the low CO2 experiments should match the measured values from the high CO2 experiments. The measured trait shifts are from chamber studies, GC1: Curtis and Wang 1998 (700 ppm), GC2: Poorter and Navas 2003 (690 ppm), GC 3: Wang et al. 2012 (700 ppm) and FACE experiments, FACE: Ainsworth and Long 2005; Ainsworth and Rogers 2007 (560 ppm). Percentage values indicate magnitude of trait shift as compared to current, levels of CO2
TraitGC1 (700 ppm)GC2 (690 ppm)GC3 (700 ppm)FACE (560 ppm)
ActualExtrapolatedActualExtrapolatedActualExtrapolatedActualExtrapolated
A max       +31% +27% ± (5)
A net +28% +63% ± (12) +28% +61% ± (12) +14% +63% ± (12) +26% +34% ± (6)
gs −11%38% ± (6)  −32%38% ± (6)−21%25% ± (4)
WUE      

+68%(C3)

+6%(C4)

+48% ± (6)
DW+28% +88% ± (21)

+48%(C3)

+12%(C4)

+86% ± (20)

+25%(C3)

−3%(C4)

+88% ± (21)   
SLA  −13%15% ± (5)−10%15% ± (5)−6%9% ± (3)

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Conflict of Interest
  10. References
  11. Supporting Information

This meta-analysis seeks to quantify and aggregate current knowledge on plant traits in low CO2. Few traits were measured for many species, and data were found for a limited number, 45, of species (Table 1). Due to the limited number of species and trait measurements, comparison between plant types, woody–herbaceous, C3–C4, was difficult. This limited data set should be taken into account when reviewing the results. Interesting results do however emerge. In response to reduced CO2, plants adjusted both physiological and morphological traits (Figure 1, Table 1). The magnitude of trait adjustment varied among species for 6 of 20 traits examined (Table 1). This suggests that on order to cope with low CO2 different species adjust different traits. This species-level heterogeneity in response though was not clearly based on simple functional groups (Figure 2).

Below, we discuss trait shifts at low CO2 moving from leaf to ecosystem scales, keeping in mind that all the results discussed below are all short-time scale, plastic responses (Gerhart and Ward 2010). Plant plastic response to changing conditions occurs at different levels of organization after different periods of time (Nicotra et al. 2010).

Leaf-level responses to low CO2

We found support for our hypotheses at the leaf level. We hypothesized that low CO2 would lead to lower photosynthetic rates (A), which leads to reduced growth, and that in order to partially ameliorate the photosynthetic rate reduction, plants would have higher leaf nitrogen and larger stomatal conductance (gs) (Sage and Reid 1992; Medlyn et al. 2011). Focussing first on A and gs, Anet substantially decreased at a 50% decrease in CO2 and gs increased considerably, indicating that the increase in stomatal conductance is not enough to keep up with lower atmospheric carbon concentration.

Water use efficiency (WUE) decreased proportionally with CO2 (following Franks et al. 2013) but did show significant variation between species. Such a strong increase in water demand suggests great consequences for plants experiencing drought stress. However, recovery from drought at low CO2 was found to be similar between C3 (Abutilon theophrasti) and C4 (Amaranthus retroflexus) plants at low CO2, due to less leaf loss and stomatal closure than expected for C3 species (Ward et al. 1999). Suggesting that there are trait shifts that mitigate some ill effects.

When comparing the magnitude of the response of A, gs and WUE at high CO2 to the responses, we found at low CO2 a contrasting picture emerges. At low CO2, no difference in WUE was found between plant types, although at high CO2, C4 plants increase their WUE less than C3 plants (Poorter and Navas 2003). For C3 plants, the response to high CO2 seems similar in magnitude as to low CO2 for WUE. For the components of WUE, at high CO2, the adjustment of Anet was greater in magnitude as to low CO2, whereas the effect on gs was comparable to low CO2, although one meta-analysis reported a far lower decrease in gs at high CO2 (Table 2). The extent to which gs can be reduced might be limited though. Leaf thermal regulation, which is impaired at very low stomatal conductance, could put a limit on the possible decrease. Paleo-evidence suggests this may have been relevant during a previous “rapid” transition to a high CO2 atmosphere at the Triassic–Jurassic boundary (Mcelwain et al. 1999).

At the level of enzymes and leaf chemical composition, we found no evidence that RuBisCO content and chlorophyll content were affected by reduced CO2; the power of our test was limited because few studies measured those traits (Table 1). Leaf nitrogen (N) content per mass did increase but N content per area decreased in response to low CO2; this contrasting response might be explained by the higher specific leaf area (SLA) in low CO2. A higher SLA indicates lower nitrogen containing mass per area. Whether or not the nitrogen increase per mass is the result of an increase in certain nitrogen-rich chemical compounds or a decrease in carbon-rich compounds, as carbon is a limiting resource, is unclear. Clearly more work is needed on the enzymatic and chemical response of plants to low CO2.

Plants adjust their leaf morphology in order to cope with low concentrations of atmospheric CO2. At half of ambient CO2, leaf SLA increased by 15%. This increase has two potential advantages to the plant: first, when CO2 is limiting, more leaf area per unit C invested in leaves allows for lower carbon costs per unit carbon capture; second, at the leaf level, higher SLA leaves may improve mesophyll conductance (Vitousek et al. 1990). While increased gs at low CO2 can maintain internal CO2 concentration up to a point, at low CO2 concentrations the diffusion of CO2 inside the leaf can become limiting to photosynthesis as well (Keenan et al. 2010). Higher SLA indicates either thinner or less dense leaves with more internal air space, which in many cases leads to greater mesophyll conductance to CO2 (Loreto et al. 1992). However, a higher SLA could also be a result of less starch or less other nonstructural carbohydrate present in the leaf at low CO2 concentrations (Poorter et al. 2009). It is interesting to find that at high CO2, SLA follows the same trend as at low CO2. The relative contributions of the above factors to reducing and increasing SLA are an interesting avenue to pursue further.

Plant-level responses to low CO2

While plants adjust their gas exchange and leaf morphology in response to reduced CO2, photosynthetic rates are nevertheless reduced, resulting in less biomass but also shifts in allocation between root and shoot. Plant biomass (g dry weight) decreases proportionally at a 50% CO2 reduction, which is more pronounced in belowground biomass as is illustrated by a reduction in root-to-shoot ratio. This could be the result of plants balancing their nutrient gain and their carbon gain to the now more limiting carbon resource (Bloom et al. 1985; Chapin et al. 1987) or some specific source-sink relationship between root and shoot, that is, fixed in the plant's metabolism, similar to the idea of a fixed ci/ca ratio for species across time (Ehleringer and Cerling 1995; Gerhart et al. 2012; Franks et al. 2013). At low CO2, photosynthesis per area is lower, so the amount of sugars available for the roots is less per unit of shoot biomass. This shift in allocation at low CO2 may have important implications for species interactions, particularly in tree–savanna grass interactions (Bond and Midgley 2012): at low CO2, the regrowth capacity following disturbance of tree species is much diminished, adding weight to the importance of fire and herbivory as ecosystem shaping factors (Kgope et al. 2010; Bond and Midgley 2012). Next to allocation, the increased SLA at low CO2 might lead to greater food availability and thus pressure from herbivores as high SLA leaves are eaten more readily (Poorter et al. 2009). Faster decomposition rates and nutrient cycling of high SLA leaves would also have large ecosystem effects by allowing faster nutrient cycling (Cornelissen et al. 1999).

Plant types, growth environments, and low CO2 response

Differences between C3, C4, woody, and herbaceous species were not as pronounced as hypothesized. Results suggested differences between plant types although the responses of only few traits were significantly different. C3 herbs had a significantly greater increase in SLA than C4 herbs, which on average showed no response. Decreases in biomass diminished starting from herbaceous C3, woody C3 to herbaceous C4. However, only the difference between woody and herbaceous C3 was significant with woody plants having a smaller reduction in biomass at low CO2. This result mirrors that at high CO2 experiments where increased CO2 generally leads to a greater relative growth rate (RGR) increase for fast growing, herbaceous, plants than slow growing, woody, plants (Poorter and Navas 2003). As fast growers “win” more at high CO2 they “lose” more at low CO2. While the smaller decrease in biomass for herbaceous C4 than for C3 is not significant, it should be noted that the average reduction in Anet and Amax is also lower than in C3 herbs. This may point to a smaller reduction in biomass accumulation for C4 herbs via a smaller reduction in photosynthesis. With more data on woody and C4 species, such differences between plant types and their underlying mechanism may become more apparent.

The results presented here summarize the effects of low CO2 at high water and high nutrients. One important caveat to consider, however, is that that there are many potential interactions between CO2, water, and nutrients as is shown by some studies. Low nutrients in the form of low P limited photosynthetic rates even further at low CO2 in Lupinus albus (Campbell and Sage 2006) and Populus deltoides (Lewis et al. 2010). C3 (Abutilon theophrasti) and C4 (Amaranthus retroflexus) plants recovered similarly from drougth at low CO2 (Ward et al. 1999). This shows that other environmental factors strongly influence the effect of CO2 on plant traits. However, most data were available for well-watered, high-nutrient growth experiments. The interactions between CO2, light, nutrients and water are clearly important when extrapolating from growth chamber experiments to glacial environments, but a full understanding of the interactions would require many more or more extensive studies.

Experimental results presented here on plastic responses need to be put into context with other sources of information on plant traits in the past including measurements on paleomaterials. In the past, CO2 has proven to be a strong selective agent altering worldwide floristic composition (McElwain et al. 2005). Thus, evolution and selection have likely occurred with increasing CO2, and it is therefore important to determine whether the traits of modern plants grown under low CO2 compare to the traits of plants that lived in a low CO2 atmosphere. The regeneration of Silene stenophylla buried in Siberian permafrost for over 30 ka (Yashina et al. 2012) provides an interesting opportunity for testing the response of modern plants and ancient plants to low CO2 and how similar they are. It is also likely that in the period since the low CO2 in the Last Glacial Maximum (LGM), species with short generation turnover have evolved more compared with long generation turnover species, although the drastic increase from 280 to 700 pmm within 250 years (1850–2100) will likely constrain the values to which traits have been adjusted. A full understanding of plant response to the transition from the LGM to current and future CO2 levels must include both evolutionary adaptation and plastic responses.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Conflict of Interest
  10. References
  11. Supporting Information

In conclusion, we found that, despite the more limited set of low CO2 studies compared with high CO2 studies, a general response is emerging. Plant response to reduced atmospheric CO2 involves a complex suite of trait adjustments. In order to diminish the effects of reduced CO2 plants open their stomata wider, invest more in aboveground biomass and increase their SLA. Despite these adjustments, photosynthetic rate is nevertheless reduced, leading to a proportional reduction in biomass accumulation. Both trait adjustment and growth effect varies among species, but this variation does not appear to be a function of simple plant functional groups. Trait adjustments at low CO2 as compared to high CO2 were proportionally similar for gs, WUE and SLA but responses at low CO2 were greater than proportional for Anet and biomass. In other words, the data suggest that in terms of water relations and leaf morphology, the responses to low and high CO2 are proportional and opposite. Carbon gain and whole plant growth rate are more complex—responses to low CO2 in these cases are more extreme. At high CO2, other factors such as nutrient and light availability could control these traits. To understand the response of plants to future high CO2, it is important to understand how and when other factors become drivers for certain traits. Our understanding of plant response to CO2 benefits from data from both low and high CO2 conditions. The shape of that response will become increasingly relevant in a high CO2 future.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Conflict of Interest
  10. References
  11. Supporting Information

This study was financially supported by Grant 142.16.3032 of the Darwin Center for Biogeosciences to R. Aerts. This is publication number DW-2013-1005 of the Darwin Center for Biogeosciences, which partially funded this project.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Conflict of Interest
  10. References
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Conflict of Interest
  10. References
  11. Supporting Information
FilenameFormatSizeDescription
ece3836-sup-0001-TableS1.xlsxapplication/msexcel24KTable S1. Overview of trait measurements reported for three or more species. V: Traits reported with error value, X: Traits reported without SE. Traits are ordered from most to least measured in low CO2 studies. SLA/LMA, specific leaf area or recalculated leaf mass per area (g·m−2); gs, stomatal conductance (mol·m−2·s−1); DW, plant dry weight (g); Amax, maximum photosynthesis (μmol·m−2·s−1); Anet, net photosynthesis (μmol·m−2·s−1); WUE, water use efficiency (mmol·mol−1); PNUE, photosynthetic nitrogen use efficiency (μmol·mmol·N−1·s−1); r/s ratio, root-to-shoot ratio DW; Ci/Ca, ratio intercellular to atmospheric CO2; % Leaf N, percentage N per weight; % Leaf mass, percentage leaf DW of total DW; Root DW, root dry weight (g); Leaf DW, leaf dry weight (g); Chlorophyll, chlorophyll content (μmol·g−1); Leaf N, N per area (g·m−2); %Stem mass, percentage stem mass of total weight; RuBisCO, RuBisCO content (g·m−2); Shoot DW, shoot dry weight (g); %Root DW, percentage root DW of total DW; Pore size, diameter of stomatal pore (μm).
ece3836-sup-0002-DataS1.csvapplication/PDF160KData S1. Meta-analysis data sheet compiled and used for this paper.
ece3836-sup-0003-DataS2.pdfapplication/PDF10KData S2. Column headers used in the meta-data file with their description and meaning.
ece3836-sup-0004-Supplement-References.docxWord document24K 

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.