Here we test an experimental approach that produces estimates of the location, timing and extent of water uptake by individual plant species in the field. A tracer injection experiment was used to estimate the proportion of active roots across soil depths. These proportions were multiplied by species-specific estimates of evapotranspiration. This approach was repeated in early- and late-season conditions in adjacent native- and non-native-dominated shrub-steppe plant communities. Resulting estimates of water uptake by depth were compared to soil water content. Both communities contained a dominant plant that relied on shallow (i.e., 0–10) soil water, a sub-dominant that relied on deep- (i.e., 90 + cm) soil water and a second sub-dominant that used late-season pulses of shallow water. Combining tracer uptake with evapotranspiration estimates indicated that non-natives removed nearly twice the amount of water early in the season as natives, especially from shallow depths. This was consistent with measures of soil moisture. However, modeled estimates of root water uptake were not always consistent with measurements of soil moisture. Despite the intensive sampling conducted in this study, future use of this approach will require more complete tracer sampling of the plant species present and improved estimation of evapotranspiration in sparse, diverse plant canopies.