The Paris Agreement underscores the need to evaluate the effects of artificial CO2 removal on climate and ecosystems. While many studies focus on temperature responses to positive and negative CO2 trends, the impact on terrestrial productivity, particularly in Europe, remains unclear. Europe's sensitivity to climate extremes and agricultural productivity changes with rising CO2 emissions is well-established. This study aims to address the knowledge gap regarding the effects of climate-carbon cycle interactions on European terrestrial productivity during net CO2 removal scenarios. Understanding this is crucial for predicting future agricultural yields and developing effective mitigation strategies. The researchers designed a comprehensive simulation using a full-complexity Earth System Model to investigate the response of Gross Primary Production (GPP) to symmetric CO2 ramp-up and ramp-down phases. The ramp-up phase simulates a 1% annual increase in CO2 from 2001 to 2140, quadrupling initial levels, followed by a symmetric 1% annual decrease from 2141 to 2280 in the ramp-down phase. This design allows for a direct comparison of the effects of increasing and decreasing CO2 concentrations on GPP. The study specifically focused on the European region given its known sensitivity to climate change and its importance in global food production.

Numerous studies have investigated the global temperature response to positive and negative CO2 concentration trends, utilizing Earth system models to address climate variability under net negative CO2 emissions. Research highlights the significant impact of the carbon cycle on climate extremes and agricultural productivity in Europe as CO2 emissions rise. However, the specific effects of climate-carbon cycle interactions on terrestrial productivity in Europe during net CO2 removal remain poorly understood, creating a critical gap in knowledge for policymakers aiming to create effective climate change mitigation strategies. This study bridges this gap by using a comprehensive simulation to analyze the detailed interaction between CO2 levels and GPP in Europe under a CO2 removal scenario.

The research employed the Community Earth System Model version 1.2 (CESM1.2) for simulations. CESM1.2 integrates the Community Atmospheric Model (CAM5), Parallel Ocean Program (POP2), Community Ice Code (CICE4), and Community Land Model (CLM4). The model has a high horizontal resolution (approximately 1°) and includes 30 vertical atmospheric levels and 60 vertical ocean levels. The land model incorporates the carbon-nitrogen cycle. Two idealized CO2 pathways were used: a constant scenario (900 years at 367 ppm CO2, representing present-day conditions) and a varying scenario (28 ensemble members). The varying scenario involved a 1% annual increase in CO2 for 140 years (ramp-up), reaching four times the initial level (1468 ppm), followed by a symmetric 1% annual decrease for 140 years (ramp-down) until reaching the initial level (367 ppm), and finally a constant CO2 concentration of 367 ppm for 220 years (restoring phase). Ensemble members differed only in their initial atmospheric and oceanic conditions. To investigate the impact of North Atlantic SST on European GPP, idealized simulations with the Atmospheric General Circulation Model (AGCM), comprising CAM5 and CLM4, were conducted. These simulations imposed uniform +1°C and -1°C SST anomalies over a specified region of the North Atlantic under present-day conditions, using 20 ensembles with varying initial conditions. In addition, multi-model simulation data from the Carbon Dioxide Removal Model Intercomparison Project (CDRMIP) were analyzed using seven Earth System Models. Observed data (SST, near-surface temperature, precipitation, cloud cover, wind, radiation, geopotential height, evaporation, soil moisture, and GPP) from various sources (NOAA, CRU, NCEP-DOE, GLEAM) were used for analysis and validation. The significance of differences between simulations was assessed using a Kolmogorov-Smirnov two-sample test.

The study revealed a striking asymmetry in the European GPP response to symmetric CO2 ramp-up and ramp-down. During the ramp-up, GPP increased, reaching 25-30 gC m⁻² month⁻¹. However, during the ramp-down, GPP declined significantly faster than expected, returning to its initial state 35% quicker than anticipated. The rapid decline is attributed to an asymmetrical response in precipitation and soil moisture. Precipitation and soil moisture decreased during the ramp-up phase and showed a significant delay in recovery during the ramp-down phase. Analysis showed that North Atlantic SST exhibited a strong asymmetric response: continuously increasing during the ramp-up phase but rapidly dropping during the ramp-down. This cooling is linked to a delayed recovery of the AMOC. This SST cooling correlates with reduced precipitation and soil moisture in Europe and consequently with the dramatic reduction in GPP. Multi-model simulations from CDRMIP supported this asymmetry. Regression analysis using observed data revealed that a cooling of the North Atlantic SST strongly correlates with decreased GPP in central and eastern Europe (15-20 g C m⁻² month⁻¹ K⁻¹), affecting Mediterranean forests significantly. Model experiments, simulating +1°C and -1°C SST anomalies in the North Atlantic, replicated the observed patterns: NASST cooling led to reduced precipitation, soil moisture, and GPP over Europe. This was linked to a dipole pattern of atmospheric pressure systems, with high pressure over Europe and low pressure in the North Atlantic. The study also found a strong positive correlation between AMOC strength and NASST during the ramp-down phase (r = 0.95), and between NASST and GPP in Europe (r = 0.62). The delayed AMOC recovery led to reduced meridional heat transport to the North Atlantic, accelerating SST cooling and consequently the decrease in European GPP. The analysis of the hysteresis of GPP and AMOC further supports this linkage.

The findings highlight the critical role of the delayed AMOC recovery and associated North Atlantic SST cooling in driving the unexpectedly rapid decline in European GPP during CO2 removal. The strong asymmetry in the GPP response underscores the nonlinearity of the climate system and the potential for unforeseen consequences of CO2 removal strategies. The study’s robust findings, supported by both single-model and multi-model analyses, have important implications for policymakers. The faster-than-expected reduction in GPP, coupled with warmer and drier conditions, increases the risk of extreme weather events (flash droughts) and severely impacts agricultural productivity. The results emphasize the importance of considering the ocean circulation dynamics and their teleconnections when planning and evaluating CO2 removal strategies.

This study demonstrates a significant asymmetry in the European GPP response to CO2 forcings, highlighting the unexpectedly rapid reduction in GPP during CO2 removal due to delayed AMOC recovery and resulting North Atlantic SST cooling. This has major implications for agricultural productivity and policymaking. Further research should investigate the sensitivity of this asymmetry to various factors, including the maximum CO2 level and the use of CO2 emission-driven scenarios for more realistic carbon cycle simulations.

The study uses idealized CO2 pathways, prescribing CO2 concentrations without considering interactions with ocean and land ecosystems. Future research should explore CO2 emission-driven scenarios to improve realism. The model’s representation of specific regional climate processes, particularly the interaction between soil moisture, precipitation, and vegetation dynamics, might warrant further investigation for improved accuracy. The study primarily focuses on European GPP, and the findings may not be directly generalizable to other regions.