Transition from positive to negative indirect CO2 effects on the vegetation carbon uptake

Terrestrial ecosystems absorb approximately 30% of anthropogenic CO2 emissions, playing a crucial role in climate change mitigation. The terrestrial carbon sink has more than doubled in the past five decades, partly due to increased vegetation carbon uptake under elevated atmospheric CO2 concentration (eCO2). eCO2 influences vegetation carbon uptake through two mechanisms: a direct effect stimulating photosynthesis and water-use efficiency (eCO2(dir)), and an indirect effect through climate change (temperature, water regime, nitrogen availability) (eCO2(ind)). Recent studies show a declining trend in eCO2(dir). Understanding the dynamics of eCO2(ind) is crucial for predicting future terrestrial carbon budgets, as its sign and temporal variation are expected to increasingly control the future trajectory. However, the dynamics of this indirect CO2 effect and its relative importance compared to the direct effect remain largely unknown, creating substantial uncertainties in the effectiveness of land-based climate mitigation policies. eCO2(ind) stems from the non-linear effects of eCO2-induced climate change on terrestrial GPP, encompassing various pathways (plants' responses to temperature, water supply, vapor pressure deficit (VPD), and their interactions). These pathways interact with eCO2(dir); for instance, rising VPD reduces stomatal aperture, modulating transpiration and the positive CO2 fertilization effect on photosynthesis. The challenge lies in quantitatively disentangling eCO2(ind), especially at regional-to-global scales, where local-scale findings from Free-air CO2 enrichment (FACE) experiments may not be applicable. While studies have reported weakening temperature-vegetation relationships, increasingly negative impacts of VPD, and increasing water constraints on vegetation growth, these focus on single climate drivers, neglecting covariation and interactions. This study addresses these knowledge gaps by investigating the dynamics of eCO2(ind) at the global scale.

Previous research has highlighted the significant role of terrestrial ecosystems in carbon sequestration, showing a more than doubling of the terrestrial carbon sink in recent decades. This enhancement is attributed, in part, to the increased carbon uptake by vegetation under elevated atmospheric CO2 concentrations (eCO2). Two primary mechanisms govern eCO2-induced changes in vegetation carbon uptake: direct effects on photosynthesis and water-use efficiency, and indirect effects through alterations in climate and environmental conditions. Studies have documented a declining trend in the direct CO2 fertilization effect, emphasizing the growing importance of understanding the indirect effects of eCO2 on the terrestrial carbon budget. However, the dynamics of the indirect effect remain poorly understood, with existing research often focusing on individual climate drivers such as temperature or water availability, neglecting the complex interplay of multiple factors. This gap in knowledge underscores the need for a comprehensive assessment that integrates multiple drivers and their interactions to accurately predict the future trajectory of the terrestrial carbon cycle.

This study investigated the dynamics of eCO2(ind) at the global scale from 1982 to 2014 using both satellite retrievals and an ensemble of Earth system models (ESMs) from CMIP6. Potential changes in eCO2(ind) up to 2100 were projected under the SSP5-8.5 scenario. Factorial simulations from fully coupled and biogeochemically coupled experiments were used to disentangle the eCO2(ind) signal. The robustness of model-based results was evaluated by retrieving eCO2(ind) from satellite observations (eCO2(ind)obs) using a climate analog framework. eCO2(dir) was derived through multiple non-linear regression, incorporating CO2 and climate drivers, exploring its relationship with eCO2(ind) across time and space. Finally, the sensitivity of eCO2(ind) to land aridity was investigated. Seven CMIP6 ESMs were used, and data were resampled to a common 0.5° × 0.5° grid. The growing season was defined based on temperature and precipitation thresholds, accounting for variations across different climate zones. Satellite-based GPP data derived from near-infrared reflectance of vegetation (NIRv) were used to validate model results. The climate analog approach identified years with similar climate but differing CO2 concentrations to disentangle direct and indirect CO2 effects from observational time series. Multiple non-linear regression models, incorporating CO2, temperature, precipitation, VPD, and cloud cover, were used to estimate eCO2(dir). The sensitivity analysis explored the relationship between eCO2(ind) changes and variations in surface soil moisture, serving as a proxy for terrestrial water availability. Statistical analyses included t-tests and Mann-Kendall tests to assess the significance of trends and changes.

Analysis of historical CMIP6 simulations (1982–2014) revealed a significant decrease in global eCO2(ind) from 0.24 ± 0.32 gC m⁻² ppm⁻¹ (1982–1996) to −0.04 ± 0.24 gC m⁻² ppm⁻¹ (2000–2014). This shift to negative eCO2(ind) was particularly prominent in cold and dry climate zones, especially boreal regions. Satellite-observed eCO2(ind)obs confirmed this global weakening effect, with an overall change of −0.38 gC m⁻² ppm⁻¹. Future projections under the SSP5-8.5 scenario indicate a further decline in eCO2(ind), with the global mean settling on negative values by 2086–2100 (−0.36 gC m⁻² ppm⁻¹ compared to 1982–1996). This decreasing signal was significant over 46.5% of global vegetated land, predominantly in the Northern Hemisphere. Analysis of eCO2(ind) on net ecosystem carbon uptake (NEP) showed a similar declining trend. Global eCO2(dir) also decreased significantly during the historical period and is projected to decline further in the future. Model-based estimates of eCO2(dir) decline were lower than those derived from satellite products and factorial experiments, possibly due to simplifying assumptions in CMIP6 models. The relative contribution of eCO2(ind) to the net effect of eCO2 (eCO2(net)) will likely decrease from 11.1% (1982–1996) to −22.6% (2086–2100), while eCO2(dir)'s contribution increases. However, the combined decline in eCO2(ind) and eCO2(dir) may result in a negative eCO2(net), with eCO2(ind) becoming the dominant driver of future GPP dynamics. Analysis of the relationship between changes in eCO2(ind) and land surface drying showed that eCO2(ind) generally declines with land drying in humid regions but weakens in water-limited conditions. This is partially attributed to CO2 and drought-related enhancements in water-use efficiency (WUE), mitigating water constraints.

This study provides strong evidence of a decline in the indirect effect of eCO2 on global vegetation carbon uptake over the past three decades, supported by both model simulations and satellite observations. The projected decline in this positive indirect effect, particularly in high-latitude regions, coupled with the concurrent decrease in the direct CO2 effect, raises concerns about the future capacity of terrestrial ecosystems to act as a significant carbon sink. The complex interplay of multiple climatic factors influencing primary productivity necessitates an integrated approach, which this study provides by considering the combined effects of temperature, precipitation, VPD, and cloud cover. The increasing water limitation, evidenced by widespread projected decline in soil moisture, emerges as a critical driver of the weakened indirect eCO2 effect. This finding is robust across different proxies of terrestrial water availability and temporal window lengths. The complex interplay of climate change and natural disturbances, such as insect outbreaks and pathogen spread, is also likely to contribute to the transition from positive to negative indirect CO2 effects, though these processes are not fully represented in current ESMs. The discrepancies between model-based and observation-based estimates of eCO2(dir) decline highlight the limitations of current ESMs in accurately capturing the complex interactions influencing the carbon cycle.

This study demonstrates a significant decline in the indirect effect of elevated atmospheric CO2 on vegetation carbon uptake, transitioning from positive to negative in recent decades, particularly in high-latitude regions. This, along with a concurrent decline in the direct CO2 effect, suggests a diminishing capacity of terrestrial ecosystems to absorb atmospheric CO2 in the future. Increasing water limitations are identified as a key driver of this trend. The limitations of current Earth system models in representing the complex interactions among climate change, disturbances, and vegetation dynamics are highlighted. Future research should focus on improving the representation of these processes in ESMs to enhance the accuracy of future carbon cycle projections and inform more effective land-based climate mitigation strategies.

While this study provides a comprehensive analysis of the transition in indirect CO2 effects, some limitations exist. The reliance on CMIP6 ESMs introduces model-specific uncertainties, and discrepancies between model-based and observation-based estimates of eCO2(dir) highlight the need for further model improvements. The study's focus on the SSP5-8.5 scenario limits the generalizability to other emission scenarios. Furthermore, while the study explores the relationship between water limitation and the weakened indirect CO2 effect, other factors such as changes in nutrient availability and the effects of biotic disturbances warrant further investigation.