Technical Comments

Comment on “Unexpected reversal of C3 versus C4 grass response to elevated CO2 during a 20-year field experiment”

See allHide authors and affiliations

Science  31 Aug 2018:
Vol. 361, Issue 6405, eaau3016
DOI: 10.1126/science.aau3016


Reich et al. (Reports, 20 April 2018, p. 317) reported that elevated carbon dioxide (eCO2) switched its effect from promoting C3 grasses to favoring C4 grasses in a long-term experiment. We argue that the authors did not appropriately elucidate the interannual climate variation as a potential mechanism for the reversal of C4-C3 biomass in response to eCO2.

Reich et al. (1) presented results of a long-term free-air CO2 enrichment experiment. The results showed that elevated CO2 (eCO2) favored C3 grasses rather than C4 grasses during the first 12 years; however, the pattern reversed during the subsequent eight years. It appears that their observations regarding the changes in C4-C3 grasses under eCO2 condition did not reflect the effects of inter-annual variations in the ambient rainfall and temperature during the 20-yr experimental period, leading to uncertainties in their results.

The effect of eCO2 on plant biomass largely depends on the ambient rainfall and temperature (2, 3). However, Reich et al. (1) found that the responses of C4 and C3 grasses to eCO2 had negligible dependence on these important climatic factors, determined by estimating the effects of multiple variables on C4-C3 biomass with repeated-measures analysis. According to the Cedar Creek weather data from Fort Snelling near the Saint Paul International Airport, the ambient total rainfall (316–722 mm) and average temperature (18.6–21.4°C) during the growing season had considerable inter-annual variations during the 20-yr experimental period. Using generalized linear models, we found that both the growing-season rainfall and average temperature positively correlated with the effect of CO2 on C4 biomass and the growing-season average temperature negatively correlated with the effect of CO2 on C3 biomass (Fig. 1). Without potential collinearity among the explanatory variables and order effects of repeated-measures analysis, our analysis is more appropriate to estimate the effect of individual variable on the response of C4 or C3 biomass to eCO2 with an accurate and interpretable predictor.

Fig. 1 Relationships between the CO2 effect on total C4-C3 biomass and growing-season climate.

The biomass data are from the measurements of Reich et al. (1). CO2 effect size = biomass under eCO2 condition – biomass under ambient CO2 condition. The Cedar Creek weather data are from Fort Snelling near the Saint Paul International Airport ( The biomass and weather data are shown as 3-yr moving averages centered over the middle of each 3-yr group. The relationships between CO2 effect size and climatic factors were analyzed using generalized linear models (CO2 effect size – temperature + rainfall; family = Gaussian; link = identity). The partial R2 of each climatic factor was obtained using the rsq. partial function with the rsq package in the R version 3.2.2.

The change in C4 biomass showed a sharp decrease from 2005 to 2008 (Fig. 2), and the C3 biomass also reached the lowest level during this period (1). Water stress during summer might have led to the decrease in biomass because summer rainfall during these dry years was about 53% less than the average of other years (Fig. 2). After these dry years, eCO2 favored C4 but not C3 grasses. Besides the asymmetric changes in net nitrogen mineralization rates between C4 and C3 soils as suggested by Reich et al. (1), we offer two other possible mechanisms for the “winner”—C4 grasses. First, increased growing-season average temperature might favor C4 than C3 grasses under eCO2 condition. The growing-season average temperature significantly increased by approximately 0.98°C before and after the dry years (Fig. 2; t = –3.6; P < 0.01). By the two-way ANOVA with CO2 (ambient CO2 versus eCO2) and average growing-season temperature (before versus after the dry years) as fixed factors to determine the effects of eCO2 and temperature on the 3-yr moving averaged C4 biomass, we found that increased growing-season temperature might interact with eCO2 to affect log10-transformed C4 biomass (F = 4.4; P < 0.05). As suggested by other studies and as shown in Fig. 1, the warm-season C4 grasses can grow better than C3 grasses under higher temperature conditions (4), and can enhance their sensitivity to eCO2 with increasing temperature when soil moisture content is not limited (46). Second, C3 grasses as cool season species lose their positive responses to eCO2 with increase in the ambient temperature as shown in Fig. 1.

Fig. 2 Inter-annual trajectories of C4 total biomass at the ambient (red) and elevated CO2 (blue) levels and summer rainfall (orange).

The Biomass data from the measurements of Reich et al. (1) are shown as 3-yr moving averages centered over the middle of each 3-yr group. The Cedar Creek weather data are from Fort Snelling near the Saint Paul International Airport.

Understanding the directions and magnitudes of responses of C4 and C3 grasses to eCO2 is crucial in modeling carbon-climate feedbacks. It is difficult to predict the changes in plant biomass dynamics in an intricate ecosystem based only on the photosynthetic pathways. Several studies have shown that the relative effects of eCO2 on the biomass of C4 and C3 grasses are highly influenced by soil water availability and temperature (26). We argue that the interpretation of the biomass data would be more meaningful by appropriately considering the effects of inter-annual variations in the ambient rainfall and temperature. Our analysis and interpretation of the biomass data provides insights different from those of Reich et al. (1), but we fully support their call for long-term experiments.

References and Notes

Acknowledgments: We thank Prof. Reich for discussing with us. The work was supported by the National Key Research and Development Program of China (2017YFC1200100), National Science Foundation of China (41630528 and 31670491), and Young Thousand Talents Program Scholar.
View Abstract

Navigate This Article