Two hypotheses are tested: 1) monitoring stations (e.g., Mauna Loa) are not able to measure changes in atmospheric concentrations of CO₂ that are generated by changes in terrestrial vegetation at distant locations; 2) changes in the atmospheric concentration of carbon dioxide do not affect terrestrial vegetation at large scales under conditions that now exist in situ, by estimating statistical models of the relationship between satellite measurements of the normalized difference vegetation index (NDVI) and the atmospheric concentration of carbon dioxide measured at Mauna Loa and Point Barrow. To go beyond simple correlations, the notion of Granger causality is used. Results indicate that the authors are able to identify locations where and months when disturbances to the terrestrial biota “Granger cause” atmospheric CO₂. The authors are also able to identify locations where and months when disturbances to the atmospheric concentration of carbon dioxide generate changes in NDVI. Together, these results provide large-scale support for a CO₂ fertilization effect and an independent empirical basis on which observations at monitoring stations can be used to test hypotheses and validate models regarding effect of the terrestrial biota on atmospheric concentrations of carbon dioxide.

KEYWORDS: Carbon dioxide; Mauna Loa; NDVI