Lithology, topography, and spatial variability of vegetation moderate fluvial erosion in the south-central Andes

• Published in: Earth and planetary science letters 2020-12, Vol.551, p.116555, Article 116555
• Main Authors: Seagren, Erin G., Schoenbohm, Lindsay M., Owen, Lewis A., Figueiredo, Paula M., Hammer, Sarah J., Rimando, Jeremy M., Wang, Yang, Bohon, Wendy
• Format: Article
• Language: English
• Subjects: 10Be terrestrial cosmogenic nuclides; Bayesian regression; Central Andes; fluvial erosion; vegetation
• ISSN: 0012-821X; 1385-013X
• DOI: 10.1016/j.epsl.2020.116555

•Spatial variability of vegetation predicts fluvial erosion in the central Andes.•Interactions of topography and lithology are secondary influences on erosion rates.•Metrics that capture distribution of controls provide better correlations than basin-wide means. Understanding how tectonics, climate and lithology interact to control fluvial erosion is complicated because these factors are spatially-variable and they may not be well-represented by mean values. We address these complications using eight new and 54 published 10Be catchment-wide fluvial erosion rates from the south-central Andes. We assess how tectonics, climate, lithology, and topography control erosion through bivariate and multivariate Bayesian regression analysis. We first compare catchment-wide mean values of independent variables compared to other summary statistics and find that metrics that capture extreme values (e.g., 90th percentile) and spatial variability (e.g., 90th minus 10th percentile) produce stronger correlations. This suggests that catchment-wide means may oversimplify the roles of tectonics, climate, and lithology in influencing erosion rates. We find that the overall variability of erosion rates in the south-central Andes is best explained by a combination of lithologic resistance and spatial variability in both vegetation (using the normalized difference vegetation index, NDVI) and topography (using specific stream power). Despite poor bivariate correlations, both lithologic resistance and spatial variability of specific stream power are significant regressors in our multivariate modeling. Lithology influences the relationship (i.e., linearity) between topography and erosion rates. Spatial variability of NDVI produces the strongest correlation with erosion rates of any of the variables we consider. Hence, spatial variability of NDVI both accounts for potential non-uniform vegetation responses to climate and also incorporates the role of both humid climates (high 90th percentile) and large bare regions (low 10th percentile) within a single catchment.