Tropical nighttime warming as a dominant driver of variability in the terrestrial carbon sink

Abstract

The terrestrial biosphere is currently a strong carbon (C) sink but may switch to a source in the 21st century as climate-driven losses exceed CO2-driven C gains, thereby accelerating global warming. Although it has long been recognized that tropical climate plays a critical role in regulating interannual climate variability, the causal link between changes in temperature and precipitation and terrestrial processes remains uncertain. Here, we combine atmospheric mass balance, remote sensing-modeled datasets of vegetation C uptake, and climate datasets to characterize the temporal variability of the terrestrial C sink and determine the dominant climate drivers of this variability. We show that the interannual variability of global land C sink has grown by 50-100% over the past 50 y. We further find that interannual land C sink variability is most strongly linked to tropical nighttime warming, likely through respiration. This apparent sensitivity of respiration to nighttime temperatures, which are projected to increase faster than global average temperatures, suggests that C stored in tropical forests may be vulnerable to future warming.

Keywords: asymmetrical warming; carbon budget; climate change; climate feedback; inversion model.

Fig. 1.

Increasing variance in atmospheric growth rate of CO₂ is driven by increasing variance of the terrestrial C sink from 1959 to 2010. Annual components of the global C budget (PgC⋅y⁻¹; shading is ±1 SD): AGR of CO₂ (A), anthropogenic E_F (B), anthropogenic E_L (C), S_O (D), and terrestrial C uptake (E). (F–J) Twenty-year moving windows of variance of each component. (K–O) Density of linear slopes of the variance for each flux across all Monte Carlo simulations, with the percentage of slopes that were positive (as shown by shading).

Fig. 2.

Interannual variability of NEE (inverted for visual clarity) is driven primarily by tropical nighttime temperature. (A) Anomalies of detrended NEE (green) and detrended tropical nighttime (red; Tmin) temperatures from 1959 to 2013. (B) Anomaly of detrended NEE compared with anomaly of detrended Tmin temperatures from 1959 to 2013. The dashed line indicates the best fit of ordinary least squares regression.

Fig. 3.

Respiration appears to mediate the impact of tropical nighttime temperature on the interannual variability of NEE. (A) Partial regression of NEE vs. tropical maximum (Tmax) temperatures, where Tmin temperatures have been removed. (B) Partial regression of NEE vs. Tmin temperatures, where Tmax temperatures have been removed. (C) Anomaly of detrended global GPP compared with the anomaly of detrended Tmin temperatures from 1982 to 2010. (D) Anomaly of detrended global respiration (R) compared with the anomaly of detrended Tmin temperatures from 1982 to 2010. The dashed line indicates the best fit of ordinary least squares regression.