LAUSR

Water is one of the main agent of erosion in many environmental settings, but erosion rates derived from beryllium-10 (10Be) suggests that a relationship between precipitation and erosion rate is statistically non-significant on a global scale. This might be because of the strong influence of other variables on erosion rate. In this global 10Be compilation, we examine if mean annual precipitation has a statistically significant secondary control on erosion rate. Our secondary variable assessment suggests a significant secondary influence of precipitation on erosion rate. This is the first time that the influence of precipitation on 10Be-derived erosion rate is recognized on global scale. In fact, in areas where slope is <200m/km (~11°), precipitation influences erosion rate as much as mean basin slope, which has been recognized as the most important variable in previous 10Be compilations. In areas where elevation is <1000m and slope is <11°, the correlation between precipitation and erosion rate improves considerably. These results also suggest that erosion rate responds to change in mean annual precipitation nonlinearly and in three regimes: 1) it increases with an increase in precipitation until ~1000 mm/yr; 2) erosion rate stabilizes at ~1000 mm/yr and decreases slightly with increased precipitation until ~2200 mm/yr; and 3) it increases again with further increases in precipitation. This complex relationship between erosion rate and mean annual precipitation is best explained by the interrelationship between mean annual precipitation and vegetation. Increased vegetation, particularly the presence of trees, is widely recognized to lower erosion rate. Our results suggest that tree cover of 40% or more reduces erosion rate enough to outweigh the direct erosive effects of increased rainfall. Thus, precipitation emerges as a stronger secondary control on erosion rate in hyper-arid areas, as well as in hyper-wet areas. In contrast, the regime between ~1000 and ~2200 mm/yr is dominated by opposing relationships where higher rainfall acts to increase erosion rate, but more water also increases vegetation/tree cover, which slows erosion. These results suggest that when interpreting the sedimentological record, high sediment fluxes are expected to occur when forests transition to grasslands/savannahs; however, aridification of grasslands or savannahs into deserts will result in lower sediment fluxes. This study also implies that anthropogenic deforestation, particularly in regions with high rainfall, can greatly increase erosion.