Isotopic (¹³C and ¹⁴C) evidence supporting the transformative effect of cattle on north Australian vegetation

Recently in this journal, Sharp & Whittaker (2003) have presented a compelling argument for the irreversible and rapid transformation of a largely treeless riverine grassland to a dense woodland on the Victoria River in the north-west of the Northern Territory in the monsoon tropics of Australia (Fig. 1). Their case was based on the evaluation of multiple hypotheses using a diverse range of correlative evidence, including analysis of sequences of historical aerial photography, vegetation survey, ethnography and dendrochronology. They concluded that the cause of the transformation was sustained overgrazing by cattle in the 1960s and 1970s. The absence of a long-term historical context was a significant gap in their argument. It is possible that such transformations are cyclical in response to ‘natural’ process such as edaphic changes in response to the development of the Victoria River floodplain. Location of the six soil profiles with carbon stable isotopes collected on the Victoria River floodplain at Bradshaw Station (a) and the change in canopy cover apparent from aerial photography taken in 1948 and 1993 (b). Radiocarbon dating and isotopic analyses of soil organic matter (SOM) accumulated in soil profiles have provided invaluable insights into tropical vegetation dynamics (e.g. Freitas et al., 2001; Pessenda et al., 2001). Tropical grasses have the C4 photosynthetic pathway that discriminates less strongly against the rare stable isotope 13C relative to trees, shrubs and herbs with the C3 photosynthetic pathway (Dawson et al., 2002). The difference in fractionation between C3 and C4 pathways can be expressed as δ13C values, defined as the difference in the ratio of 13C to 12C in a sample compared with that in the Vienna-PDB international standard: C3 plants have δ13C values that vary between −32 and −22‰, whereas C4 plants have values that vary between −17 and −9‰ (Boutton et al., 1998). We use the stable carbon isotope composition of SOM to provide a historical context to the recently published assertion of Sharp & Whittaker (2003) that overgrazing by cattle caused the conversion of treeless plains to woodland. Radiocarbon dates from SOM at the base of six soil profiles were determined to provide geochronological control. Aerial photographs taken in 1948 and 1993 were used to identify six sites on the Victoria River floodplain with different levels of woody vegetation thickening (Fig. 1) following the procedure outlined by Sharp & Whittaker (2003). The geo-coordinates of the six sites were determined using a geographical information system (GIS). A global positioning system (GPS) was used to locate these sites in the field. A hand-held auger was used to collect 400 g soil samples from every 10 cm to 1.4 m depth. The soil samples were subsequently air-dried, weighed and sieved to remove the gravel fraction (> 2000 μm). The total concentration of carbon and nitrogen and the carbon stable isotope composition of SOM were determined from bulked soil fraction (< 2000 μm) following the methods described in Bowman & Cook (2002). For each of the basal samples from the six sites, wet sieving was used to recover particles of less than 25 μm diameter to sample the slow turnover and passive carbon pools (e.g. Bird & Poisai, 1997; Boutton et al., 1998). These are most likely to yield reliable radiocarbon ages of SOM that has gradually accumulated in the soil profile through various pedogenic processes including eluviation and bioturbation. A 5.0–50.0 mg portion of the fine fraction was sent to the ANTARES AMS facility (Lawson et al., 2000) for radiocarbon dating. Prior to analysis these samples were treated with HCl (2 m) at 60 °C for 3 h to remove possible carbonates, washed with deionised water, then dried at 60 °C. AMS dating of the six soil profiles revealed that the age of the floodplain sediments at 1.4 m depth varied between 3000 years to modern age (Fig. 2). Sediment age was not related to proximity to the current river channel suggesting that the floodplain expanded periodically rather than through the gradual and uniform deposition of alluvium (Fig. 1). The sandy loam soils showed limited evidence of pedogenesis consistent with their recent deposition. The concentration of total carbon was low (mean 1.13, range 0.6–2.2%) throughout the profile although concentrations generally declined with depth (r2 = 0.316) (Fig. 2). Total nitrogen also declined with soil depth (r2 = 0.631) and was strongly correlated with carbon concentrations (r2 = 0.727), with a mean C : N ratio of 17.95. For each profile, the δ13C stable carbon isotope signal has remained between −18‰ and −10‰, consistent with that of a tropical grassland dominated system and closely match the unpublished values obtained by Cook (2001) in his study in the Victoria River District of stable nitrogen in savanna vegetation and soils. Cook (unpubl. data) found that a savanna with a basal area of 2 m ha−1 had mean stable carbon isotope composition of −15‰ and a range of −18.2‰ to −12.8‰. These δ13C values are much higher than those reported for Eucalyptus tall grass savanna on laterite soils with a basal area of between 6 and 15 m ha−1 which range between −17 and −26‰ (Bowman & Cook, 2002; Bowman et al., 2004). These data to support Sharp & Whittaker's (2003) conclusion that the conversion of the grassland to woody vegetation, with a basal area of >10 m2 ha−1, was a unique and recent event given that it has no parallel in the last 3000 years of the environmental history of the floodplain. Concentration of carbon and carbon stable isotope composition (δ13C value) in soil organic matter from six soil profiles on the floodplain. The radiocarbon age at the base of each profile is also shown. The ANSTO laboratory codes for the AMS dates are as follows: A = OZG130; B = OZG131; C = OZG132; D = OZG133; E = OZG1304; F = OZG371. It remains unclear if the source of the SOM is from the vegetation growing on the floodplain that has gradually accumulated in the soil profile through various pedogenic processes. It is possible that some of the SOM is derived from soils higher up the Victoria River Catchment, thereby confounding this simple interpretation. Nonetheless, if the SOM is derived from upstream, the results are still consistent with Sharp and Whittaker's conclusion because the repeat photographic record throughout this region shows that prior to the advent of cattle grazing 100 years ago the currently heavily wooded riverine plains were treeless and dominated by tall tropical grasses with a C4 photosynthetic pathway (Jacklyn, 2000; Lewis, 2002). The Department of Defence permitted our research at Bradshaw Field Training Area. We thank Tom Vigilante for help with the fieldwork and laboratory work, Dan McIntyre who prepared the map, Garry Cook for providing unpublished information an arranging the isotopic analyses, Lynda Prior for providing invaluable assistance with the data analysis and preparation of the manuscript, and Don Franklin for his comments on the paper. The AMS dating was supported by AINSE grant 02/007 and the fieldwork was supported by a Northern Territory University Project Grant. Professor David Bowman is Director of the Australian Research Council Key Centre for Tropical Wildlife Management at the Charles Darwin University. In collaboration with a numerous colleagues, he uses a diverse range of tools including remote sensing and geographical information analyses, stable isotopes, modelling, biological survey and molecular analyses to understand the evolution and history of Australian landscapes at a variety of spatial and temporal scales. Dr Ben Sharp currently works as a Senior Scientist for the New Zealand Ministry of Fisheries. His previous research examined ecological processes driving landscape-scale vegetation dynamism in north Australian savanna landscapes. Professional interests include ecosystem-based fisheries management, landscape ecology, and the application of traditional ecological knowledge to modern conservation management. Dr Ugo Zoppi is a research scientist at the Australian Nuclear Science and Technology Organisation. By means of radiocarbon measurements carried out using the ANTARES (Australian National Tandem for Applied RESearch) accelerator, he contributes to numerous research programs in the archaeological and environmental sciences.