Anthony Dosseto

Materials Contributed through SERC-hosted Projects

Other Contributions (2)

Time evolution of soils part of Vignettes:Vignette Collection
Soils are vital for the sustainability of ecosystems and human societies (e.g. (Montgomery, 2007)), and therefore it is important to understand how their rates of production and erosion are balanced. Furthermore, the rate of soil erosion is strongly linked to water salinity and can dramatically affect drainage systems (e.g. (Chhabra, 1996)). Previously, cosmogenic isotopes, such as in-situ beryllium-10 (10Be), have been used to determine rates of soil production (Heimsath et al., 1997). However, this approach requires the assumption that soil erosion and production are balanced (that the soil thickness is in steady-state) and it has only been possible to test this hypothesis in a few instances (Heimsath et al., 2000). Moreover, because soil production and erosion rates are assumed to be equal, it is impossible to use this approach to evaluate whether the soil is aggrading or degrading. Recently, uranium-series (U-series) isotopes have been proposed as an independent means to determine the rates of both soil and saprolite production. By comparing soil production rates inferred from U-series isotopes to erosion rates derived from 10Be, albeit is possible to quantitatively evaluate soil evolution. For a system that has remained closed for more than 1 Ma (such as bedrock) all of the uranium-series nuclides will be in secular equilibrium (i.e. a daughter-parent activity ratio equal to unity) (Bourdon et al., 2003). Because the relative mobility of radionuclides during weathering is believed to be uranium-234 (234U) > uranium-238 (238U) >> thorium-230 (230Th), the residues of recent weathering (e.g. soils) deviate from secular equilibrium and are anticipated to have (234U/238U) < 1 and (230Th/238U) > 1 (Chabaux et al., 2003; Dosseto et al., 2008a) (parentheses denote activity ratios). The amount of disequilibrium depends on both the extent and age of weathering processes. For example, 238U-234U and 234U-230Th can record weathering events up to 1 million years (Ma) and 300 thousand years (ka) old, respectively, because 234U and 230Th, the daughter nuclides of each system, have half-lives of 245 and 75 ka, respectively. In a study of a lateritic profile in the Brazilian Amazon basin, Mathieu et al. (1995) have shown that the time required to produce a 15m thick weathering profile is about 300,000 years, implying a migration rate of the weathering front of ~ 50 mm/ka. In a lateritic profile from Burkina-Faso, Dequincey et al. (2002) measured 238U-234U-230Th compositions which were difficult to reconcile with the common assumption that the relative radionuclide mobility during weathering is 234U>238U>>230Th. Nevertheless, they developed a model to estimate that the profile was developed over a timescale > 300 ka. In a recent study, Dosseto et al. (2008b) have investigated soils of moderate thickness (< 1m), developed in a temperate climate. They selected a site in southeastern Australia where soil erosion rates have been previously determined using cosmogenic isotopes (Heimsath et al., 2000). At this site, Heimsath et al. (2000) also showed that soil thickness is likely to be in steady-state and soil erosion rates could be used to infer rates of soil production. Soils are relatively thin, but they overlay a thick saprolite horizon (20-30 m; (Green et al., 2006)) developed over a granodiorite. Dosseto et al. (2008b) measured significant radioactive disequilibrium between 238U-234U and 234U-230Th in saprolite and soil material. They used these results to model that it takes 0.55 to 6.2 Ma to develop the thick saprolite unit, which is equivalent to saprolite production rates of 4 to 46 mm/ka. The average residence time of material in the soil was also constrained and ranges from 8 to 38 ka. Dosseto et al. investigated several models possible to calculate soil production rates and derived values ranging from 13 to 59 mm/ka. These values are comparable to the soil erosion rates obtained using cosmogenic isotopes. Thus, the combination of U-series and cosmogenic isotope techniques provide an independent mean to assess any potential imbalance between soil loss (erosion) and gain (production from the saprolite or bedrock). At the site studied in southeastern Australia, devoid of any human activity, it was shown that there is no major imbalance and that the soil thickness is in steady-state. Future studies should look at sites that have undergone severe soil loss as a result of intensive land use.

Links between climate change, vegetation and erosion in Australia during the last 100,000 years part of Vignettes:Vignette Collection
In an era where climate is changing rapidly under the influence of human activity, there is an increasing need to understand how soil and water resources will cope with these changes. While past variations in climate are well documented, the ways in which soil and water resources have responded to these past climatic changes is less well understood. To address this question, one needs to be able to quantify how soil erosion and sediment transport have varied in the past and compare these variations to the climatic record. However, it is only recently that such a quantification has become possible, with the development of cosmogenic nuclides (von Blanckenburg, 2006) and uranium-series (U-series) isotopes (Dosseto et al., 2008a). These techniques offer the opportunity to re-construct past variations in soil erosion and sediment transport, and by comparing these variations to climatic and vegetational records, to investigate the links between erosion, and climatic and biological changes in Australian environments. In a recent study, DePaolo et al. (2006) suggested that the measurement of radioactive disequilibrium between 238U and 234U nuclides could be used to trace physical weathering: when 238U decays close enough to the surface of a grain (typically less than 0.03 um), the daughter product, 234Th, can be ejected from the grain before decaying into 234U (Fig. 2). Thus, a depletion of 234U compared to 238U occurs at the surface of the grain. However, this depletion can only be measured if the surface area/volume ratio of the grain is large enough, which is the case for grains with a diameter typically smaller than 50 um. The 238U -234U radioactive system will deviate significantly from secular equilibrium only when a grain is reduced to a size less than 50 um. Hence, by measuring the (234U/238U) activity ratio of fine sediments, it is possible to date when the bedrock was physically weathered into a fine sediment. This is termed comminution age. In sediment deposits, the comminution age integrates the duration of storage of sediments in soils, transport in the river and deposition (Fig. 1). Thus, if the deposition age of sediments can be determined (for instance by luminescence dating), by subtracting the deposition age from the comminution age, then the length of time between sediment formation to deposition can be determined. This is termed the residence timeof sediments in the catchment. Thus, for the first time it is possible to re-construct past variations in sediment residence time by determining the comminution age of sediments with different deposition ages. Dosseto et al. (in review; 2008b) have used this approach to investigate how erosion has responded to climate variability in southeastern Australia during the past 100,000 years. Palaeo-channels from the Murrumbidgee River (Fig. 2) have been sampled. The deposition age of sediments collected in these channels ranges from < 50 years to 100,000 years. The Murrumbidgee River is a 900 km-long tributary of the Murray River that, along with the Darling River, forms the largest river system in Australia (drainage area ~ 1,000,000 km2). The Murrumbidgee River flows westward from the Snowy Mountains across the Riverine Plain before joining the Murray River, the largest river system in Australia (Fig. 2). The measurement of uranium isotopes of these sediments yields comminution age ranging from 42 thousand years (kyr) for sediments deposited during the Last Glacial Maxium (LGM) to 540 kyr for sediments deposited to an interglacial 100 kyr ago. The inferred palaeo-residence times range from 27 kyr (LGM sediments) to 480 kyr (modern sediments). Thus, we observe that during LGM, sediments spend an amount of time in the catchment an order of magnitude smaller than during an interglacial period like today or 100 kyr ago. To explain these variations Dosseto et al. (in review; 2008b) compared them to different records of climatic and hydrologic parameters (Fig. 3). Interestingly, variations in sediment residence time seem to match some climatic parameters, such as the atmospheric carbon dioxide (CO2) content as measured in ice cores, or sea-surface temperatures, suggesting a link between changes in climatic conditions and erosion in the Murrumbidgee River basin. However, whilst we would expect a direct link between river discharge and sediment residence time (the more water in the river, the more efficient erosion, the shorter the residence time), there is not a good match between variations in sediment residence time and those in palaeo-discharge, reconstructed using the morphology of palaeo-channels. Dosseto et al. suggest that variations in sediment residence time values reflect a change in the provenance of sediments: during LGM, sediments originate from the headwater regions, where young soils are eroded; while during interglacials, sediments carried in the river mostly derive from re-mobilization of old alluvial sediments, stored in the Riverine Plain for tens of thousands of years. This hypothesis is supported by the co-variations observed between palaeo-residence time values and the type of vegetation in the upper catchment: during LGM, the upper catchment was mostly covered with shrubs, which would have little effect in reducing erosion and sediments eroded from young soils could reach the river. During interglacials, the upper catchment is covered with woodlands. Tree roots contribute to significantly decrease erosion in the headwaters such that little erosion would occur there. Instead, the sediment budget of the river would be dominated by bank erosion in the alluvial plain, as it is observed for the modern river channel (Wallbrink et al., 1998). These results suggest that during the last glacial cycle, changes in sediment dynamics have been controlled by catchment vegetation dynamics. This implies that climate may impact erosion indirectly by dictating the type and density of vegetation coverage in the catchment. This emphasizes the role of living organisms on landscape evolution (Dietrich and Perron, 2006), illustrated here over a glacial cycle.