Over the past four decades, multiple experiments have been performed to estimate the response of plants to higher concentrations of carbon dioxide (CO2
) under field conditions. Agricultural crops have been of particular interest due to the strong concerns for future food security and safety [1
] and to explore the potential advantages of the fact that rising CO2
may stimulate plant growth. The Intergovernmental Panel on Climate Change [2
] projected that CO2
concentrations are likely to be in the interval 420–1300 ppm (RCP2.6 and RCP8.5, respectively) by the year 2100. Consequently, to assess possible yield stimulations, there is a need to estimate crop responses to elevated CO2
) over a range of concentrations, although single experiments mostly used only one or sometimes two levels of eCO2
Wheat (Triticum aestivum
Linnaeus) is one of the most studied crops regarding eCO2
responses, since it is one of the major food crops globally. Plant growth is generally stimulated by eCO2
, resulting in higher yields [3
]. The growth stimulation is a result of both enhanced photosynthesis (C3 crops), but also improved water use efficiency (C3 and C4 crops) due to reduced stomatal conductance [4
]. Short-term plant responses to eCO2
usually include a higher net CO2
assimilation, while downregulation of photosynthesis can occur over longer time scales (growing season) [5
]. The CO2
fertilization effect on C3 photosynthesis will mainly occur until the concentration is saturated at ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco), which is not the case at the current atmospheric concentration (400 ppm). In addition, the water saving effect can significantly improve plant performance [6
]. However, it remains uncertain how these effects translate into crop yield responses over a wider range of CO2
concentrations under field conditions.
Enclosure systems, such as open top chambers (OTC), have been widely used in CO2
field experiments, but also questioned since they alter the micro-climate of the plants and thus may modify the magnitude of crop responses to eCO2
]. Comparison of conditions in OTCs to the open field show that temperatures and vapor pressure deficits are higher inside chambers and airflow is altered in the plant canopy [9
]. The use of OTCs will also reduce transmission of solar radiation and shift the ratio between diffuse and total radiation. The field tunnels (e.g., Rawson [11
]) used in some eCO2
experiments with crops are likely to alter the micro-climate in a similar manner as OTCs. Furthermore, it is questionable whether results from experiments with plants rooted in pots can be extrapolated to field conditions. Potted plants have a restricted rooting volume that may affect the response to eCO2
]. At the same time, plants grown in pots are likely to experience a higher, light interception, since they are usually not surrounded by a closed canopy, which may exaggerate effects. Free Air CO2
Enrichment (FACE) systems have been developed to create a less artificial experimental setup compared to enclosure systems like OTCs and tunnels. On the other hand, FACE systems have the drawback of not being able to reach strongly elevated concentrations for eCO2
treatments (no experiments using CO2
concentrations above 600 ppm) and possibly less stable concentration levels that may lead to underestimation of plant CO2
In addition to grain yield as such, there are a number of yield variables of both agronomical and economical importance for grain yield, which are critical to study in order to understand how eCO2
affects the growth pattern of crops. In the present study we included the following yield components and aspects of grain physical characteristics: harvest index, grain number, grain mass, and specific grain mass. Harvest index represents the fraction of the total aboveground biomass found in the harvestable products at maturity, which is central in crop breeding as a measure of the efficiency with which resources (solar radiation, water, and fertilizers) are used and converted into the desired harvested plant component. Grain mass (equivalent to 1000-grain weight) and specific grain mass (volume weight or test weight) are quality aspects that affect the market price of wheat grain. Higher values of these variables are related to a larger flour yield, while low values indicate small and malformed grains of poor quality [14
]. Historical improvements in wheat grain yield has been positively correlated to an increase in grain number per unit area [16
]. There is, however, a trade-off between increasing the number of grains and grain mass if photosynthetic rates remain unchanged [17
]. The CO2
-induced grain yield (mass per unit area) stimulation can be a result of increased grain number (per unit area) and/or average grain mass.
effect on wheat grain yield was reviewed by Amthor [18
], where response functions showed that studies performed in laboratory chambers and greenhouses compared to field experiments had almost doubled yield stimulation per increased ppm in the range of 350–750 ppm. There was, however, only one FACE experiment conducted at that time. Using meta-analysis, Wang, Feng, and Schjoerring [4
] estimated the overall CO2
impacts on wheat crop physiology and yield, showing an average yield stimulation of 24%, and an effect of similar magnitude was estimated by van der Kooi et al. [19
]. In line with Amthor [18
], Wang, Feng and Schjoerring [4
] found differences in yield response between enclosure systems, where closed-top chambers had a yield stimulation close to 40%, significantly higher than all other types of exposure systems (OTCs, FACE, and greenhouses). They also found that the grain yield stimulation by eCO2
was significantly stronger in potted plants compared to field grown; however, not taking into account that there is an association between rooting environment and the enclosure system used. Studies with greenhouses and closed-top-chambers mainly used pots, while plants in FACE experiments were grown in field soil and OTC studies use both potted and field rooted plants.
Our study aims to provide an up-to-date summary of eCO2
effects on wheat crop yield, based on observations from ecologically realistic field experiments and excluding treatments with additional environmental stress, such as heat, drought, elevated ozone, or low nitrogen (N) supply. In addition, we will consider CO2
effects on various wheat yield components (harvest index, total aboveground biomass, grain mass, grain number, and specific grain mass), which was not included by Amthor [18
]. There has also been discussions regarding the relationship between crop CO2
response and environmental conditions controlling site productivity [14
], such as growing season temperature and water availability, which varies considerably between different regions of wheat cultivation. Bishop, Leakey, and Ainsworth [20
] concluded that growing season temperature was not a good predictor of yield CO2
response but found a negative relationship between water supply (precipitation + irrigation) and yield CO2
response; however, the dataset was rather limited for this analysis since not many experiments reported total water supply. To address this, we examined the relationship between the eCO2
effect on grain yield and the absolute grain yield in control treatments, used as a proxy for wheat crop productivity in the area where the experiment was performed. In the present study, we used meta-analysis to estimate the overall magnitude and statistical significance of responses to eCO2
treatments. Response functions were derived in order to understand the gradual change over the range of CO2
concentrations used in the experiments, and to test if the response with increasing CO2
was linear or non-linear. By these approaches our study aims to answer the following research questions:
Does yield stimulation by eCO2 saturate at high CO2 concentration?
Is eCO2 yield stimulation dependent on experimental conditions (fumigation technique, rooting environment, wheat type, geographic region)?
Is there a link between eCO2 yield stimulation and agronomic productivity?
Are yield components (total aboveground biomass, harvest index, grain mass, grain number, specific grain mass) equally affected by eCO2?
presents the average effect of eCO2
on a range of wheat yield variables, using aCO2
as the reference. Grain yield significantly increased by 25.6% (CI 20.9–28.5%) under eCO2
and total aboveground biomass showed an equal response of 24.8% (CI 21.7–28.1%). Grain number was significantly enhanced by 22.3% (CI 17.6–27.1%) due to eCO2
, while grain mass showed a small, but still significant increase by 2.1% (CI 0.6–3.7%). Harvest index and specific grain mass remained unaffected under CO2
enrichment, effects estimated to 0.4% (CI −1.4–2.2%) and −1.6% (CI −8.8–3.3%), respectively.
The response function for the relationship between grain yield and CO2
concentration (Figure 2
a) showed a strong non-linear relationship (R2
= 0.44). Grain yield increased with higher CO2
to reach a maximum yield response at ~600 ppm. Details for all regression models, linear and quadratic, are presented in Table 1
. Figure 2
b gives the relationship between the grain yield CO2
response (response ratio −1%) with ΔCO2
(difference in CO2
concentration between aCO2
treatment in each experiment), demonstrating that there was no relationship between grain yield stimulation and difference in concentration between control and elevated treatments.
Meta-analysis for subgroups of grain yield data (Figure 2
c) showed an overall similar response for the different groups. There were no significant differences between exposure systems (FACE, OTC, and tunnels), but an indication of a smaller eCO2
effect in the tunnel systems (13.1.0%, CI 5.5–22.4%) compared to the 24.1% yield stimulation in OTC (CI 19.9–28.5%) and 28.2% in FACE (CI 20.1–36.9%).
There was no significant difference in terms of grain yield between the two levels of eCO2 treatment tested. The average yield response for plants grown in CO2 concentration < 600 ppm was 26.5% (CI 21.1–32.3%) and 22.3% (CI 17.1–27.6%) for concentrations > 600 ppm. The comparison was also made for the subset of OTC data to avoid the potential influence of different types of exposure systems using different levels of eCO2 treatment. Results from OTCs showed no significant differences between eCO2 levels above and below 600 ppm, with an eCO2 effect of 23.9 (CI 18.4–29.7%) and 24.4% (CI 18.0–31.6%), respectively. The latter observation (OTC < 600 ppm) was also within the same range as grain yield response in FACE systems, thus when comparing FACE and OTC using the same level of eCO2 there were no significant differences in yield stimulation.
Comparison of the rooting environment showed an almost identical response, where grain yield under eCO2 increased by 25.2% (CI 19.6–31.1%) for potted plants and 24.5% (CI 19.8–29.5%) for plants rooted in field soil. Since potted plants were only used in OTC experiments, this dataset was also compared to a subset of yield data restricted to plants grown in OTC and field soil, where the eCO2 effect was 23.4% (CI 17.6–29.6%), also very similar to the effect for pot grown plants. The average eCO2 concentration in potted plants was 625 ppm, which was comparable to the field grown plants in OTCs with an average of 620 ppm.
Categorizing experiments by region revealed that eCO2 effects on grain yield were significantly stronger in experiments performed in Asia (40.0%, CI 31.0–49.8%) compared to North America (16.0%, CI 10.0–23.1%) and Europe (22.1%, CI 18.0–26.3%), while the Australian experiments (29.3%, CI 20.0–39.3%) did not significantly differ from any of the other regions (North America p = 0.29, Europe p = 0.12, and Asia p = 0.39). It should, however, be noted that the number of observations is small for North America and Asia, with four and five comparisons, respectively. The average eCO2 effect on grain yield in winter wheat was 22.5% (CI 16.9–29.4%), while the response in spring wheat was stronger (25.4%, CI 21.0–30.0%), but with no significant difference between the two wheat types.
a shows the relationship between the grain yield CO2
response and absolute grain yield in aCO2
, representing the agronomic productivity of the site and year. Data were categorized by exposure system and rooting environment (Figure 3
b), concentration level of eCO2
treatment (Figure 3
c), and region (Figure 3
d). Whatever subdivision of data that was made, the relationship was very similar in all cases, indicating that the estimated average effect of the agronomic productivity on the eCO2
response was robust. Quadratic fit for the complete dataset (Figure 3
a) gave a marginally better fit (R2
= 0.27) compared to the linear model (R2
= 0.24) and model performance was equal when considering AICc. Coefficients and model performance for all regression models are presented in Table 2
. The relationship between absolute grain yield response and absolute grain yield in aCO2
was also tested but did not show any association (R2
The relationship between CO2
concentration and total aboveground biomass (Figure 4
a) showed a strong positive non-linear relationship (R2
= 0.54) and the response function for grain number (Figure 4
d) had a similar response pattern (R2
= 0.47). Similar to the response function for grain yield, also for total aboveground biomass and grain number, the stimulation by higher CO2
gradually declined and reached a maximum response at ~600 ppm. The response function for grain mass and harvest index (Figure 4
b,c) showed weak non-significant relationships with CO2
concentration, with an R2
of 0.15 and 0.087, respectively.
Based on the observation that the response pattern for the grain number was similar to grain yield, the correlation of these responses was tested. Figure 5
demonstrates the eCO2
effect on grain number and the corresponding effect on grain yield, where the correlation of effects was very strong (r
= 0.82). The broken line (1:1-line) represents the theoretical situation where the effect of eCO2
on grain yield is entirely explained by the effect on grain number. The slope of the regression line (y
= 6.58 + 0.90x
) was not significantly different from the 1:1-line (p