Travel and residence times are well-known descriptors of hydrologic and solute transport in the vadose zone. It has been observed that their probability density functions are stationary only under specific conditions, rarely encountered in natural catchments. This study aims at demonstrating the emergence of non-stationary solute transport in a highly monitored system, and identifying the factors controlling the variations of the observed solute travel-times. 2-meters deep weighing lysimeters are exposed to stochastic rainfall sequences. Multiple derivatives of difluorobenzoate compounds are sequentially injected at different times in the system, and are analyzed in the drainage flux at the bottom outlet and at different depth within the soil profiles. Willow trees planted in the systems create a stochastic soil water deficit by evapotranspiration. As each tracer injected is analytically differentiable from the others, the computation of the tracer breakthrough curves at the lysimeter outlet allows measuring the solute travel-time distributions conditional on the injection time. The observed breakthrough curves display a large variability, emphasizing the effects of the initial conditions at the injection time and the subsequent states encountered in the system on solute transport. Two types of climate have been simulated on the lysimeters. With the precision load cells installed under each lysimeter and the water content probes deployed in the soil profiles, a detailed comparison of the water balance and storage dynamics and their influence on solute transport timing can be done.