¹⁰Be constrains the sediment sources and sediment yields to the Great Barrier Reef from the tropical Barron River catchment, Queensland, Australia

Abstract

Estimates of long-term, background sediment generation rates place current and future sediment fluxes to the Great Barrier Reef in context. Without reliable estimates of sediment generation rates and without identification of the sources of sediment delivered to the reef prior to European settlement (c. 1850), determining the necessity and effectiveness of contemporary landscape management efforts is difficult. Here, using the ~ 2100-km² Barron River catchment in Queensland, Australia, as a test case, we use in situ-produced ¹⁰Be to derive sediment generation rate estimates and use in situ and meteoric ¹⁰Be to identify the source of that sediment, which enters the Coral Sea near Cairns. Previous model-based calculations suggested that background sediment yields were up to an order of magnitude lower than contemporary sediment yields. In contrast, in situ ¹⁰Be data indicate that background (43 t km^− 2 y^− 1) and contemporary sediment yields (~ 45 t km^− 2 y^− 1) for the Barron River are similar. These data suggest that the reef became established in a sediment flux similar to what it receives today. Since western agricultural practices increased erosion rates, large amounts of sediment mobilized from hillslopes during the last century are probably stored in Queensland catchments and will eventually be transported to the coast, most likely in flows triggered by rare but powerful tropical cyclones that were more common before European settlement and may increase in strength as climate change warms the south Pacific Ocean. In situ and meteoric ¹⁰Be concentrations of Coral Sea beach sand near Cairns are similar to those in rivers on the Atherton Tablelands, suggesting that most sediment is derived from the extensive, low-gradient uplands rather than the steep, more rapidly eroding but beach proximal escarpment.

Introduction

The Great Barrier Reef (GBR) is an important cultural, economic, and environmental resource, the management of which is a high priority for the Australian government (Neil et al., 2002, Gordon, 2007, Hunter and Walton, 2008). The effect of European land use change since c. 1850 on surface water quality, specifically increased erosion and the presumed consequent increase in sediment delivery to the reef, has attracted significant scientific and management attention over the last several decades (Larcombe and Woolfe, 1999, Neil et al., 2002, McCulloch et al., 2003, McLaughlin et al., 2003, Macdonald et al., 2005). Modeling suggests that delivery of sediment from upland catchments to the reef has increased since European settlement (e.g., Neil et al., 2002, Brodie et al., 2003, Kroon et al., 2010, Kroon et al., 2012). Such sediment and associated nutrients and pesticides can be a significant pollutant when exported in large quantities from cleared lands through rivers and estuaries to the reef, increasing the nutrient load and the turbidity of coastal waters, thus reducing light penetration and photosynthesis and forming mud banks (Wolanski and Spagnol, 2000, Lambrecht et al., 2010).

Responding to concerns of pollution from catchment runoff, the Australian and Queensland governments developed the Reef Water Quality Protection Plan in 2003, and updated it in 2009 and 2013 (Reef Water Quality Protection Plan, 2013). The 2013 update calls for reducing anthropogenic total suspended sediment (TSS) loads 20% by the year 2018. However, realizing these goals is not straightforward. In order to estimate the anthropogenic TSS loads, estimates must be reasonable of the contemporary TSS loads and of the background/long-term (hereafter termed pre-European) TSS loads that are subtracted from the contemporary estimates (Kroon et al., 2012). In order to determine the significance of sediment pollution and to design cost-effective land management strategies, an understanding of how contemporary sediment loads compare to pre-European sediment loads is needed. Currently, the pre-European sediment yield estimates are based on land use models' replete with assumptions (Brodie et al., 2009).

Contemporary sediment yields of catchments draining to the GBR, based on model results and scant data, are estimated to be higher than the natural background rates (e.g., Prosser et al., 2001, McCulloch et al., 2003, Joo et al., 2005, Herr and Kuhnert, 2007). The model results are from the SedNet and ANNEX models, which predict the long-term average annual rates of erosion based on rainfall, soils, topography, crops, and land management practices (Lane et al., 1988). The SedNet model was developed in the western USA with data from gently sloping croplands and was later adapted for Australian conditions (Prosser et al., 2001). While the sediment yield predicted by SedNet for catchments at the 100,000-km² scale compared well with that determined by recent empirical studies, discrepancies were found at smaller scales of < 30,000 km² (Dougall et al., 2005). These disparities are a concern considering only two of the catchments that drain to the GBR are significantly > 30,000 km².

The contemporary, area-weighted average sediment yield for all GBR catchments (Fig. 1), as determined by several independent estimates, is between 29 and 66 t km^− 2 y^− 1 (Table 1). The estimates differ because of the number and magnitude of flood events per wet season, the concentration of suspended sediment during floods, and the land use specified for SedNet and ANNEX models (e.g., Moss et al., 1992, Furnas, 2003). Using data from one study (Brodie et al., 2009) for context, the contemporary sediment yields of the individual GBR catchments ranged from 16 to 250 t km^− 2 y^− 1. By comparison, several estimates of contemporary sediment yield for the Barron River catchment that we studied (Fig. 1) ranged between 14 and 210 t km^− 2 y^− 1 (Table 2). A recent estimate of Barron River contemporary sediment yield using the annual TSS data from 2006 to 2009 is 92 t km^− 2 y^− 1 (Joo et al., 2012), while most published estimates for the contemporary sediment yield are between 44 and 48 t km^− 2 y^− 1.

SedNet models have also been used to estimate the pre-European TSS yields that most studies reference (NLWRA, 2001, Brodie et al., 2003). To determine background sediment delivery rates, the model uses underlying assumptions that cannot be verified for pre-European conditions, such as natural vegetation cover, the absence of gullies (Brodie et al., 2003, McKergow et al., 2005), or adjusting contemporary values downward by a factor of 10 for cropping land use and by a factor of 2 to 4 for grazing land uses (Neil and Yu, 1996). Gully processes are particularly significant, as gully age, hillslope delivery ratio (connectivity of hillslopes to channels for sediment transport), and bankfull discharge and recurrence interval are the most sensitive parameters for SedNet (Fentie et al., 2005); such poor constraints on pre-European conditions could introduce large errors to the model results (Brodie et al., 2009). For the Barron River catchment, the background sediment yield estimates are between ~ 2 and 18 t km^− 2 y^− 1 (Table 3), similar to the estimates for the average pre-European sediment yield from all GBR catchments (Table 1). These estimates are equivalent to < 1 to 6.7 mm ky^− 1 of erosion and are remarkably low in comparison to the global data set for all catchment long-term erosion rates (Portenga and Bierman, 2011).

Evidence from elemental ratios (Ba/Ca) in reef coral suggests that fine sediment delivery to the reef increased after 1850 (McCulloch et al., 2003, Lewis et al., 2012). The Ba/Ca ratios from the Burdekin River show that sediment delivery from the catchment to the GBR was low from 1760 to 1850 and then increased with European settlement in the catchment (McCulloch et al., 2003), suggesting that at least fine sediment transport to the reef increased after land use change. Similarly, Ba/Ca and Y/Ca ratios in the Whitsunday Islands suggest that sediment delivery increased shortly after European settlement, probably the result of land clearing in the Proserpine and O'Connell River catchments for grazing and cropping (Lewis et al., 2012). However, the magnitude of Ba/Ca decreased after 1917, suggesting that either removal of surface sediment and revegetation of crop lands caused an overall reduction in fine sediment erosion, transport, and delivery or that the number of tropical cyclones decreased (Nott et al., 2007, Callaghan and Power, 2010). Conversely, other studies suggest that over the longer temporal scale, geological evidence shows that many of the inshore reefs have always existed in highly turbid environments and were initiated on muddy substrates (e.g., Larcombe and Woolfe, 1999, Perry et al., 2008, Perry et al., 2009, Perry et al., 2011, Perry et al., 2012).

One method to determine long-term erosion rates, analysis of in situ-produced cosmogenic ¹⁰Be in river sand (Brown et al., 1995, Bierman and Stieg, 1996, Granger et al., 1996), estimates sediment generation rates on millennial time scales directly and thus provides much needed data to compare with the contemporary estimates of sediment yield from models and from suspended sediment measurements (Neil and Yu, 1996, Prosser et al., 2001, Gordon, 2007, Herr and Kuhnert, 2007). In situ ¹⁰Be is produced by cosmic-ray bombardment at and near Earth's surface (Lal and Peters, 1967, Bierman and Nichols, 2004). Because most ¹⁰Be atoms are produced within the uppermost ~ 2 m of soil and rock, this technique integrates over ~ 10,000 years at erosion rates of 100 mm ky^− 1 and over longer time spans at lower erosion rates. An estimate of pre-European erosion rates can be made by measuring the average ¹⁰Be concentration of channel sediment sourced from across the catchment (Bierman and Nichols, 2004, Vanacker et al., 2007). Because ¹⁰Be measures erosion integrated over multimillennial time scales, most, if not all, sediment that is produced is exported from the catchment and therefore ¹⁰Be-derived erosion rates are equivalent to long-term, average, background sediment yields (Kirchner et al., 2001, Schaller et al., 2001, Lu et al., 2005).

In addition to understanding the quantity of sediment generated, identifying the sources of sediment is also useful. Such knowledge has the potential to help land managers select, at a catchment scale, which management strategies provide the greatest reduction in sediment yield at the least cost (Star et al., 2013, vanGrieken et al., 2013). Several studies have delineated contemporary sediment sources to GBR catchments using SedNet modeling and sediment tracing with conflicting results (e.g., Cogle et al., 2000, Brodie et al., 2003, McKergow et al., 2005, Wilkinson et al., 2009, Wilkinson et al., 2013, Thorburn and Wilkinson, 2012). Recent spatial modeling of catchment sediment budgets using SedNet has suggested that 70% of sediment delivered to the GBR lagoon is sourced from 20% of the catchment area, mainly within 80 km of the coast (McKergow et al., 2005, Wilkinson et al., 2009). Such modeling has also indicated that 60–70% of sediment delivered to the GBR is derived from sheetwash and rill erosion on hillslopes (McKergow et al., 2005, Thorburn and Wilkinson, 2012). However, the SedNet model predictions of sediment sources have received little evaluation against independent data in the GBR catchments. Recent sediment tracing in the Burdekin River catchment suggests that between 77 and 89% of sediment is sourced from the subsurface (Wilkinson et al., 2013), and empirical data for the Normanby catchment suggests that the SedNet model does not estimate contemporary sediment sources well (Brooks et al., 2013).

Determining the sediment sources within catchments over longer temporal scales is equally important. Such knowledge allows for focused efforts to mitigate erosion and sediment yield to levels resembling long-term, background rates to which ecosystems are adapted. In order to determine long-term sediment sources (e.g., Clapp et al., 2002, Reusser and Bierman, 2010) and catchment sediment budgets (Perg et al., 2003, Nichols et al., 2005) over millennial scales, we use in situ and meteoric ¹⁰Be as a tracer to fingerprint sediment from different areas of the Barron River catchment. The concentration of meteoric ¹⁰Be, like in situ ¹⁰Be, depends on the style and rate of land surface erosion, slower eroding surfaces will have higher concentrations of meteoric ¹⁰Be while more rapidly eroding (or more deeply eroded) land will have much lower concentrations (Pavich et al., 1985). The resulting difference in ¹⁰Be inventories caused by different rates and styles of erosion can be used to trace the source of sediment (e.g., Clapp et al., 2002, Reusser and Bierman, 2010). Thus, for both methods, by assuming that any change in ¹⁰Be during fluvial transport is negligible (rapid fluvial sediment transport relative to hillslope residence time and minimal sediment storage) nuclide inventories measured in samples collected down a drainage basin are useful for determining sediment sources and the efficacy of sediment mixing.

In this paper, we use the dual ¹⁰Be isotope analysis, in situ-produced and meteoric ¹⁰Be, to determine sediment sources and sediment generation rates in the Barron River catchment, Queensland, Australia (Fig. 1), and to provide the first nonmodel estimates of pre-European rates of sediment generation. This study, of one catchment, provides baseline sediment yield data and a proof of concept for tracking sediment in the Wet Tropics region of Queensland, a landscape that discharges sediment to the inner shelf of the Great Barrier Reef (Barron and Haynes, 2009).

Similar to many other catchments in the Wet Tropics draining to the GBR, the quartz-rich Barron River catchment is characterized by low-slope uplands and coastal lowlands separated by a steep, abrupt coastal escarpment ~ 300 m high. The uplands have a mean elevation of 581 m and a mean slope of 8.5°, while the escarpment catchments have a mean elevation of 395 m and a mean slope of 16° (Table 4). Within the catchment, land use includes intensive agriculture, recreation, tourism, conservation, urban development, and forested areas that are part of the Wet Tropics World Heritage Area (Anderson et al., 1993). Mean annual precipitation is highest in the middle catchment (~ 2500 mm y^− 1) and lowest in the upper catchment (~ 900 mm y^− 1; http://www.barronrivercatchment.org.au). The wet season spanning from November to May produces all of the major discharge events on the Barron River (http://watermonitoring.derm.qld.gov.au/host.htm).

The GBR is divided into three distinct sections. The inner shelf is dominated by granitic-derived fluvial detritus (silicoclastic and aluminosilicate clay and silt; Belperio, 1983) from adjacent catchments. The middle shelf is mostly starved of terrestrial sediments, and the outer shelf is dominated by biogenic carbonate sediments (Mathews et al., 2007). Thus, the effect of the terrestrial sediment delivered to the reef is not uniform. A recent study estimated that since the mid-twentieth century, the inner shelf region of northern Queensland experienced deposition from terrestrial sediment plumes 92 times, the middle shelf 34 times, and the outer shelf 11 times (Walsh et al., 2004); thus, at least fine sediment can travel up to tens of kilometers offshore (Ryan et al., 2007, Radke et al., 2010).