Native forest deforestation has been identified as one of the main land cover changes affecting flood risk specially during small and moderate storm events. In this regard, forest protection and reforestation are considered a nature-based solution (NbS) for flood regulation. However, there is a lack of knowledge about the effects of different deforestation spatial patterns over floods. Effects of land cover changes on floods in a humid tropical basin within the Ecuadorian Amazon are assessed distinguishing forest location and forest fragmentation. The hydrological distributed model TETIS was applied to simulate the hydrological response of a basin to extreme storms having return periods of 1, 10 and 100 years, considering five land cover scenarios. The model was calibrated and validated using nine storm samples collected at a gauge station during the years 2018 and 2020. The simulated overland flow in hillslopes and stormflows within the river channel were analyzed to i) assess the statistical differences among all land use scenarios with the Kruskal-Wallis test; ii) assess the statistical differences among pairs of both location and fragmentation scenarios through the post-hoc evaluation Dunn test; iii) assess the statistical differences in relation to the baseline. Obtained results indicate that stormflow is less sensitive than overland flow to land cover changes. Forest location have more influence than forest fragmentation over both, overland flow and storm flows. Deforestation of the upper basin represents the worst scenario for flood regulation, thus protection of existing forest, as well as reforestation of deforested areas located in the upper watersheds is a priority for flood risk mitigation and forest conservation. The results enhance our understanding of ecosystem services provided by tropical Andean foothills forests.