Factors controlling storm-generated suspended-sediment loads and concentrations from 1989-95 were examined for three watersheds in Puerto Rico of contrasting land use and human disturbance. They included a slightly disturbed 3.26 km² forested watershed (Rio Icacos), a moderately disturbed 8.39 km² pasture and forested watershed (Quebrada Blanca), and a heavily disturbed 19.4 km² urbanized watershed (Rio Piedras). Using 13 independent variables of runoff and rainfall, forward stepwise regression and factor analysis were used to determine the significant controls on suspended-sediment load and discharge-weighted sediment concentration. The 13 independent variables were classified in three time categories: (1) characteristics of the previous event, (2) characteristics between events, and (3) characteristics of the current event.

     In the moderately disturbed Quebrada Blanca watershed, which is dominated by pasture, characteristics of the current event were significant. In the more heavily urbanized and disturbed Rio Piedras Watershed, where sediment sources are more widespread, previous event and between-event characteristics were significant with between-event characteristics flushing sediment from the system. Hysteresis curves for each sampled event reflect the differences in sediment availability in each watershed. In the forested Rio Icacos watershed, 73% of the hysteresis curves are clockwise or show depletion of sediment over time. In Quebrada Blanca, predominately pasture and forest cover, 51% of hysteresis curves are clockwise and 49% are counterclockwise. In the disturbed Rio Piedras watershed, 70% of hysteresis curves are counterclockwise, indicating increasing sediment supply over time. Hysteresis curves appear strongly related to land-use types and may be a useful tool for linking sediment loads to disturbance. Development of this tool will require a better understanding of how the relative importance of different sediment sources (such as, channel versus hillslope) change during the course of a storm.