Meeting Paris Agreement targets increasingly points to a need for carbon dioxide removal (CDR) from the atmosphere. While CDR seems promising, the associated uncertainties are not well known. One unexplored aspect is whether the uncertainties in CDR and transient climate response to cumulative CO2 emissions (TCRE) are coupled to any degree, e.g. some CDR methods could have lower carbon removal potential if TCRE are high. In this paper, we quantify the coupled uncertainties between TCRE and the CDR of ocean alkalinity enhancement (OAE) and reforestation. To do so, we used a perturbed parameter ensemble with the University of Victoria Earth system climate model. These results were fed to train Kernel Flows-based model emulators, which in turn were used, together with observational data, to carry out Markov Chain Monte Carlo simulations to characterize Bayesian posterior distributions. It turns out that OAE results in slow but steady CDR, with temperature reduction nearly linearly related to the amount of carbon removed. CDR with OAE is also only weakly dependent on uncertain climate system parameters and TCRE, indicating that OAE is a reliable CDR method with respect to the coupled uncertainties between its removal potential and TCRE. CDR of reforestation slows down after 50 years, and its temperature reduction from biogeochemical effects is offset by the warming biogeophysical effects, as reforesting pastures lowers albedo. Moreover, CDR by reforestation shows a strong dependency on uncertain climate system parameters and negatively correlates with TCRE, implying that it is less reliable as a CDR method. We argue that CDR potential, timescale of CDR, and uncertainty coupling between TCRE and CDR should all be considered when designing robust climate change mitigation portfolios.