Variability in ENSO-induced carbon flux patterns

El Niño-Southern Oscillation (ENSO) is not only a driver of global carbon cycle variability, but it also provides several mechanisms of predictability. Although most Earth system models (ESMs) can reproduce the relationship between ENSO and atmospheric CO2 concentrations, the question remains whether the ESMs agree on the origins of these ENSO-related GPP anomalies. We analyse the patterns of ENSO-induced GPP anomalies in 17 ESMs to determine from which regions these GPP anomalies come from, and whether the differences among the models are driven by climate forcing or biochemistry. While most of the GPP anomalies originate from Southeast Asia and northern South America, there are large deviations among the ESMs. The combined GPP anomaly of these two regions ranges between 26% and 75% of the global anomaly among the ESMs. To find out what causes the differences, we examined two major drivers of the GPP anomalies: the size of the ENSO-induced climate anomalies, and the sensitivity of GPP to climate. On the global average, ENSO-induced climate anomalies and GPP sensitivity have similar uncertainty among the ESMs, contributing equally to the variations in ENSO-induced GPP anomaly patterns. This analysis reveals model biases in teleconnection patterns and biochemistry. Addressing these biases is a tangible goal for model developers to decrease the uncertainty in the reproduction of the global carbon cycle, and to increase its predictability.