Increasing climate variability, especially longer drought phases, shapes forest water consumption and constrains vegetation growth. However, the effects of drought-induced changes on evapotranspiration (ET) fluxes vary due to species-specific differences or features of the drought events. Here, we analyzed the seasonal variations in ET in pine, beech, and mixed forest stands in northeast Germany during the periods 2012-2021 and explored the ability of a process-based ecohydrological model (EcH2O) in reproducing the water balance components at this large-scale lysimeter site. To better understand how individual climate variables control ET fluxes, we performed simulation experiments with average climate inputs. Multi-variable calibration showed that the model reproduced well in-situ soil moisture, seepage, and interception (EI) in the three forest stands. Lower ET rates were found in the beech stand with a seasonal leaf cycle compared to those in the pine and mixed pine-beech stands with year-round foliage. This was mainly related to higher EI in the pine stand, particularly during the winter season, while transpiration (T) rates were similar. ET fluxes were above average during wet years and below average during dry years. Precipitation (P) was clearly the main driver of ET anomalies. However, only small reductions in ET were observed during the dry year 2018. This could be attributed to high P in the previous year, i.e., P legacy effects, which led to only small reductions or even positive anomalies in T. Moreover, model testing at such data-rich sites will also be valuable for other hydrological models to increase process-consistency and reliability, particularly when aiming at investigating effects of different vegetation cover. Our findings provide evidence for the strategic selection of broadleaf species in forest management practices to enhance groundwater recharge and promote sustainable water management.