The role of unpaved roads as active source areas of precipitation excess in small watersheds drained by ephemeral streams in the Northeastern Caribbean

Quantitative understanding of the impacts of land development on runoff generation is vital for managing aquatic habitats. Although unpaved roads are broadly recognized as significant sources of sediment within managed forested landscapes, their role in altering runoff response is characteristically dependent on rainstorm and watershed size. Here we evaluate the role of unpaved roads in the development of Horton overland flow and their potential to influence the delivery of runoff from small watersheds (∼1skm2) drained by ephemeral streams flowing toward coral reef bearing waters of the Northeastern Caribbean. Infiltration capacity curves for undisturbed forest soils and unpaved roads were developed based on hydrologic characterization performed with a Guelph permeameter. Results demonstrate that infiltration capacities from unpaved roads are roughly a quarter of those for forest soils. Consequently, localized precipitation excess is about four times greater on unpaved roads than on forest soils. Analyses indicate that unpaved roads generate precipitation excess roughly ten times more frequently than watershed-scale storm flow generated by the combined effects of precipitation excess and saturation overland flow. Comparison of unpaved road precipitation excess with observed watershed discharge suggests that road networks may produce localized surface runoff equal to 62% of total watershed discharge for rainstorms up to 3.0cm, and this holds even for watersheds with low and moderate road densities (0.8–2.3kmkm−2). For watersheds with high road densities (∼7.6kmkm−2), roads may contribute about one-quarter of storm flow for rain events up to 10cm. Our results stress the high sensitivity of runoff response in dry tropical watersheds to land disturbance, even when this disturbance occurs on only about 1% of the land surface. In this particular case study, unpaved roads prove capable of altering the time distribution of runoff and, by extension, sediment delivery, from one that is naturally infrequent and sporadic to one that is potentially chronic.