Landslides, hurricanes, and sediment sourcing impact basin-scale erosion estimates in Luquillo, Puerto Rico

Accurately inferring erosion rates from cosmogenic isotope concentrations in river sand assumes temporally steady concentrations; few studies test this assumption. Following Hurricane Maria in Puerto Rico, we quantified temporal variability in meteoric and in situ Be-10 (Be-10(m), Be-10(i)) on sand-sized grains of riverine transported material in landslide-prone basins. We analyzed 20 samples collected over 18 months from the channels of two nested watersheds: the Icacos (3.14 km(2), 0.09% active landslide area) and Guaba basins (0.11 km(2), 1.23% active landslide area). Be-10(i) concentrations in Icacos basin sediment remained steady over time whereas concentrations in Guaba basin sediment were initially half those in the Icacos basin and increased linearly over 18 months, constraining recovery time to \u3c2 yrs for this basin. Be-10(m) concentrations in both drainages did not change consistently over time and were not related to precipitation events; Pb-210(ex) and Cs-137 were below detection limits in all samples. Our data demonstrate that Be-10(i) concentrations in river sand can be lowered for months to years after major landscape disturbing events, such as large or extensive mass movements. Sampling soon after a landslide will result in over-estimates of long-term erosion rates. Such bias can be reduced by repeated sampling over time and by sampling numerous similar watersheds of different sizes and different concentrations of landslides in a study area. (C) 2021 Elsevier B.V. All rights reserved