<p>Exploitation of water resources beyond groundwater recharge rates has the potential to reduce baseflow to streams and impact ecosystem functioning. Areas with high biodiversity hotspots, such as wetlands and estuarine lakes, are particularly vulnerable but can also yield differences in headwater recharge estimates based on the method used. The Verlorenvlei estuarine lake (RAMSAR #525), north of Cape Town, South Africa, is a sensitive biodiversity hotspot with a high dependence on groundwater but features limited hydroclimatic and hydrogeological data. Recharge estimations from a conceptual J2000 rainfall-runoff model were used as the upper boundary inputs for a hydrogeological MODFLOW model. Recharge coefficients derived from MODFLOW were estimated to range between 0.3 and 7.0% of precipitation in the valley and 7.0–11.7% in the mountains from 2010 to 2016, where high model sensitivity was shown in the headwaters and lower primary aquifer. In comparison, recharge from the J2000 model for the same period ranged between 0.8 and 5.1% in the valley and 22.6–25.1% in the mountains, with a high sensitivity to the rate of interflow. Both J2000 and MODFLOW showed high sensitivity for the outflow rate/hydraulic conductivity for the alluvial aquifer and low sensitivity to simulated percolation/recharge, even though they were calibrated with two independent datasets (streamflow vs. groundwater levels). Coupling isotope tracers could improve the representation of slow and fast flow paths needed to constrain secondary aquifer dynamics. Results highlight the need for multiple groundwater recharge quantification methods. Combining outputs from a rainfall-runoff model as inputs into a groundwater model can reduce the data requirements for simulating groundwater recharge.</p>