Multi-element signatures of stream sediments and sources under moderate to low flow conditions

Abstract

This study assesses a simple sediment source tracing method using major- (Al, Ca, Fe, K, Mg, Mn, Na, P, Si, Ti) and trace-element (Ba, Be, Ce, Co, Cr, Mo, Nd, Pb, Sr, Th, V, Y, Zn) signatures of stream suspended particulate matter (SPM), bed sediments and soils in a small agricultural catchment in NE Scotland. Whilst most erosion studies characterise the large amounts of material mobilised at the highest flows, this study aimed to assess properties of sediments during moderate to low flow periods. These occur more frequent than intense storms and are important in linking stream sediments, near-channel sources and aquatic ecosystem impacts. Data were transformed by multivariate statistical methods to compare elemental signatures of SPM (ranging from 3 to 53 mg L⁻¹ in the stream) and stream bed sediments with a limited number of near-channel source soils. Increased concentrations of Ce, Nd, Th and Y in subsoils contributed to the ability to discriminate between surface fieldslope and stream bank erosion sources. Stream bed sediments showed close matches with compositions of stream bank and arable surface soils, but signatures of SPM differed greatly from any of the sources. Large concentrations of Cr, Pb and Zn in SPM, particularly during summer (677, 177 and 661 mg kg⁻¹, respectively) exceeded water quality standards and were linked to an accumulation of trace elements associated with biological material. The potential for within-stream alteration of SPM in relation to erosion sources was confirmed by changes in the nature of the SPM organic matter observed by IR spectroscopy. Thus the potential is shown for multi-element signatures to give information on catchment sediment sources to aid land management decisions, given careful consideration of the effects of in-stream alteration of eroded material. However, this combined information may be beneficial to process understanding linking land use and stream ecosystems at critical ecological periods.

Introduction

Soil erosion is a substantive environmental problem for soil loss, contaminant transport and for its ecological impacts on receiving waters (UNESCO, 1983, Quinton and Catt, 2007, Stutter et al., 2007). In intensively managed catchments this erosion process provides a pathway for the transport of metals, organic contaminants and faecal-associated microorganisms which partition with solid phases rather than into solution (Allan, 1986, House et al., 1998, Warren et al., 2003). Maximum concentrations and large proportions of annual sediment loads are associated with infrequent, large runoff events in catchments. However, there is an increasing understanding that more frequent, lower intensity events can transport material that may contribute disproportionately to some reactive, or contaminant components of the sediment budget (Quinton et al., 2001, Quinton and Catt, 2007). The progressive transport of material towards stream channels over successive smaller erosion events can result in particle fractionation as fines are preferentially transported and transient storage of material in riparian areas, or the channel itself (Russell et al., 2001).

The source is a fundamental parameter controlling chemical and physical properties of the sediment, contaminant transport potential, and the extent and timescales of material transport to water courses. Therefore, the management of soil erosion and particulate-bound contaminants in agricultural ecosystems requires an understanding of the nature and location of potential source materials. Sediment fingerprinting is a method currently under development to determine the sources of eroded material to watercourses (Walling, 2005) and has supplemented conventional ground surveying assessment of soil loss (e.g. Haigh, 1977, Walling et al., 1993 showed that no single parameter could reliably discriminate sources due to complex physico-chemical sorting of material during erosion. Since then, modern applications of sediment fingerprinting have used composite signatures involving a wide range of determinants. Often large and complex combinations of parameters are employed, for example physical (particle size, surface area), mineral–magnetic, chemical (acid, oxalate and dithionite extractable metals) and radiometric (Pb and Cs isotopes) properties, which may be extremely costly and time constraining analytically (Collins et al., 1997, Collins and Walling, 2002, Walling, 2005). These then often require correction factors for the effects of differing organic matter and surface area on determinant concentrations between sources and sediments (Collins et al., 1997, Russell et al., 2001). Multivariate statistical methods and mixing models are applied to these properties to determine their potential to discriminate different source materials and their proportional contributions to stream sediment loads (Yu and Oldfield, 1989, Walling et al., 1999).

The fingerprinting process can also yield important spatial information on the nature of the source material and alterations in the stream (Yu and Oldfield, 1989, Russell et al., 2001). This includes comparisons between source and watercourse bed sediments and suspended particulate material, the selectivity of chemical properties in the erosion process, erosion under different land uses (Walling et al., 1999) and temporal changes in sedimentation (Owens et al., 1999). One particular focus for sediment research is the delivery of P from agricultural land to watercourses where this, often limiting, nutrient is an important control on ecosystem trophic status (Edwards and Withers, 2007). Particulate bound P is often much more prevalent in surface waters than dissolved P forms (Stutter et al., 2008a), but this has complex relations with: (i) erosion controls such as precipitation and stream flow giving spatio-temporal variability in particulate P transport, (ii) the P status of different source materials under different soil management, and (iii) biogeochemical cycling during erosion and in the stream itself (Stutter et al., 2007, Stutter et al., 2008a, Stutter et al., 2008b). Variation in the timing and form of the sediment and associated P delivery, or in residence time in the watercourse will affect the potential for eutrophication of surface waters. Data from the fingerprinting process can thus provide valuable assessment of the links between the chemical status of the source materials, the potential impacts on the aquatic ecosystem and the means of ameliorating them.

This study aims to assess the application of a wide suite of major and trace elements, rare earth metals, nutrients and particle size data in the characterisation of stream suspended particulate matter (SPM) and bed sediments and their potential source soils in a small agricultural catchment in NE Scotland. A simple sediment tracing method is presented using total element concentrations after dissolution of sample solids. Hence, the total elemental signatures encompass the fractionation of minerals during the erosion process and any subsequent cycles of deposition and redistribution on land and in the stream. It is thought that metal compositions following total sample digestion will provide a more conservative tracing approach than using reagents which bring about selective dissolution of surface-bound elements (e.g. acid leaching). A secondary aim was to explore the linkage between the catchment near-channel soils and stream network during spring to autumn periods of moderate to low flows, a period critical to stream ecology. Most erosion studies focus on characterising the large amounts of material mobilised at the highest flows (Walling et al., 1999, Russell et al., 2001, Stutter et al., 2008b). However, this work aimed at assessing properties of sediments during more frequent, less intensive hydrologic events that are important in linking stream sediments, near-channel sources and aquatic ecosystem impacts (Quinton et al., 2001).