Abstract Reservoir sedimentation poses a critical threat to water storage capacity globally, particularly in the U.S. Great Plains where storage loss has continually declined over the past several decades. While regional management often prioritizes streambank stabilization, effectively targeting mitigation requires distinguishing between channel‐derived and upland sediment sources across varying flow regimes. This study employed sediment fingerprinting to apportion streambank, cropland and grassland contributions to the Cottonwood River and the downstream John Redmond Reservoir, utilizing a multi‐method sampling approach to capture different temporal scales of sediment transport. For flows contained within the river channel, time‐integrated traps identified streambanks as the dominant sediment source (~50%), whereas discrete storm samples underestimated bank contributions (~35%). Statistical analysis of flow regimes and trap positioning indicates this discrepancy is not driven by vertical stratification of the sediment load or particle size differences but rather by the ability of time‐integrated samplers to capture pulses of sediment transport that discrete sampling misses. In contrast, lakebed deposits in John Redmond Reservoir, which represent a longer‐term record of sediment transported by large‐magnitude floods, were dominated by cropland sources (~40%). The event‐driven shifts in sediment provenance suggest a hydro‐geomorphic threshold: during extreme, out‐of‐bank flows, hydrologic connectivity expands beyond the channel to the cultivated floodplain, mobilizing vast quantities of surficial cropland soil in exceedance of the river's bank‐dominated internal load. Consequently, watershed management faces a dual challenge: streambank stabilization is essential to reduce the chronic sediment pulses delivered during routine within‐bank flows, but soil conservation on croplands is equally critical to mitigate the large‐scale sediment delivery associated with extreme out‐of‐bank floods.
Layzell et al. (Wed,) studied this question.