Mountains and highlands, the world's natural water towers, contribute about 32% of global water discharge ranging from 20-50% in humid regions to 50-90% in semi-arid and arid regions. These water towers face heterogeneous land use and land cover (LULC) changes causing disproportionate hydrological impacts and distinct vulnerabilities. Such changes threaten water availability and increase climate-related water extremes. Understanding the hotspot patterns is critical to managing these vulnerable systems. This study investigates LULC changes at Mt. Kenya water tower and its surroundings at multiple spatial scales and evaluates their impacts on hydrology using the SWAT+ model. LULC classifications were prepared using the random forest algorithm on Landsat 7 and Landsat 8 imagery. Results showed significant overall LULC change: agriculture and mixed forest increased by +8.1% and +4.7%, while evergreen forest and shrubland decreased by -3.5% and -7.5%, respectively. Sub-basin analysis indicated heterogeneous LULC changes. The SWAT+ model showed a Kling-Gupta efficiency (KGE) of 0.80 and 0.78, and a very good percent bias (PBIAS) of -7.4 and -6.9, for calibration and validation, respectively. Significant sub-basin changes were observed in evapotranspiration (+19.6% to -7.3), water yield (+11.4% to -7.4%), surface runoff (+11.4% to -7.4%), and percolation (+13.9% to -8.3%). Correlations showed that the loss of vegetation reduced evapotranspiration and increased water yield, surface runoff, and percolation, with opposite trends for increasing vegetation. Hence, anthropogenically induced LULC change significantly impacts water towers' hydrology. The uneven impacts reveal unique LULC change patterns for which befitting adaptive and mitigative measures are needed.