A knowledge gap in our understanding of the global carbon and water cycle, which in turn impacts climate science insights and earth systems model development, is the relative dominance of the drivers of vegetation productivity and subsequently the atmospheric carbon-dioxide growth rates (CGR). There is a lively debate in the literature based on correlation or bivariate causal methods, on whether the primary driver(s) is (are) temperature (T) or water availability, specifically, precipitation (P), soil moisture (SM), and/or terrestrial water storage (TWS). Our data-driven causal analyses, which rely on Granger Causality (GC) and the Peter and Clarke Momentary Conditional Independence (PCMCI) tests, consider all potential causal variables holistically, and reveal two primary atmospheric drivers for net ecosystem exchange and CGR: vapor pressure deficit (VPD) and photosynthetically active radiation (PAR). Our findings challenge recent literature, underscore the critical role of evaporative demand and vegetation dynamics in atmospheric carbon sensitivity, and highlight the need for more robust causal testing methods.