Rock magnetism and geochemistry of two plio–pleistocene Chinese loess–palaeosol sequences—implications for quantitative palaeoprecipitation reconstruction

Abstract

Stratigraphic and spatial variations in the magnetic properties of Chinese loess are widely accepted as reflecting changes in the intensity of soil forming processes over time and space. Recently, several workers have used statistical correlations between the magnetic susceptibility and related parameters of modern loess plateau soils and modern climate to generate transfer functions to reconstruct past temporal and spatial trends in precipitation. Here we try to assess the viability of this approach by conducting magnetic and geochemical measurements of two long (∼2.5 My) loess–palaeosol sequences from different locations on the modern precipitation gradient across the loess plateau. One site, Duanjiapo, is located on the southern humid margin of the plateau; and the other site, Luochuan, is located in a drier modern environment some 160 km further north. Our results indicate discrepancies between magnetic and geochemical indicators of weathering and soil forming intensity and that any attempt to predict palaeoprecipitation at Duanjiapo using published approaches would be significantly in error. Overall our results indicate a complex response of magnetic mineral transformations to climate, including the production of both ferrimagnetic and antiferromagnetic minerals within the same climatic interval, together with the destruction of magnetic material within some of the palaeosols. Therefore we suggest considerable caution should be exercised in generating quantitative climatic reconstructions from a single magnetic parameter.

Introduction

The loess plateau in north China consists of a thick (∼100–300 m) accumulation of aeolian dust derived from arid source areas to the north and northwest. The stratigraphic sequence comprises alternating loess and palaeosol layers representing deposition over the last ∼2.5 My. It is underlain by a less visibly differentiated ‘red clay’ deposit, also of aeolian origin, representing sediment deposition extending back to ∼7 My and probably considerably earlier (e.g. Ding et al., 1998, Sun et al., 1998, Guo et al., 2002). Numerous studies of the loess–palaeosol sequence have used variations in lithology, geochemistry, grain-size and other properties to infer fluctuations in the intensity of the Asian summer and winter monsoons. In contrast with the relatively unweathered loess layers, the palaeosols exhibit darker colours, finer grain-sizes and geochemical signatures indicative of weathering and soil development. These characteristics are interpreted as reflecting warmer and more humid climates resulting from increases in the intensity of the East Asian summer monsoon (e.g. Kukla, 1987, Derbyshire et al., 1995, An, 2000).

The nomenclature of the loess–palaeosol sequence used herein follows Rutter et al. (1991): loess layers are designated ‘L’, palaeosol layers are designated ‘S’, and the layers are numbered sequentially downwards. For example, S0 refers to the youngest, early Holocene, palaeosol and L1 refers to the loess deposited during the last glacial cycle. The components of soil complexes are designated as S(i)S(i) and S(i)L(i): for example, S5S1 is the youngest soil within the S5 pedocomplex.

Soil iron oxide mineralogy is sensitive to climate and can rapidly be characterised using rock (or ‘mineral’) magnetic techniques (e.g. Thompson and Oldfield, 1986, Evans and Heller, 2003). As a result, magnetic measurements of the loess–palaeosol sequence are a popular tool for assessing variations in the degree of palaeosol development, and hence of summer monsoon intensity. The magnetic mineral transformations that take place during soil development on the loess plateau are complex; however, a key process is undoubtedly the production of significant quantities of ultra fine-grained (< 40 nm) ferrimagnetic minerals during the warmer and wetter climatic intervals which generate the palaeosols, as was first proposed by Zhou et al. (1990), Maher and Thompson (1992) and Liu et al. (1993). Maher and co-workers (e.g. Maher, 1998, Maher et al., 2003) argue convincingly for the importance of the metabolic action of iron-reducing bacteria in the production of extra-cellular magnetite grains spanning the size range from superparamagnetic to stable single domain. Irrespective of their origin, the ultra fine-grained ferrimagnets make a large contribution to the magnetic susceptibility, which can be rapidly and easily measured in the field and laboratory. Following observations of statistically significant relationships between the magnetic properties–principally the magnetic susceptibility–of modern loess plateau soils and modern climatic parameters, several workers have tried to extend the use of magnetic measurements as a quantitative proxy for reconstructing palaeoprecipitation (e.g. Heller et al., 1993, Maher and Thompson, 1994, Maher et al., 1994, Liu et al., 1995). If confirmed, such quantitative climatic reconstructions for the loess plateau area could be used to validate the predictions of global and regional numerical climate models. There is broad agreement amongst various magnetic-based palaeoprecipitation estimates for the last glacial–interglacial cycle in the central part of the loess plateau (Evans and Heller, 2001). However, several workers have reported inconsistencies between the magnitude of coeval magnetic susceptibility peaks at sites in the northern and central (with a drier modern climate) and southern (more humid) loess plateau areas. For example, in the case of palaeosol S8 in the Duanjiapo section on the humid southern margin of the loess plateau, Guo et al. (2001) found no correlation between magnetic susceptibility and the degree of pedogenesis. In addition, Han et al. (1996) noted distinctly non-linear relationships between climatic parameters and modern soils in a large set of modern Chinese soils. They also suggested that in the case of well-developed palaeosols on the southern margin of the loess plateau that the inferred very warm and humid conditions may either have favoured the production of iron oxides other than ferrimagnetic minerals or have resulted in their breakdown. Therefore, in the context of the increasing use that is being made of the magnetic properties of Chinese loess for quantitative climatic reconstruction it seems important to further investigate their temporal and spatial consistency. To this end we use a combination of high stratigraphic resolution magnetic measurements and partial geochemical analyses from two ∼2.5 My loess sequences located in contrasting modern climate regimes in order to assess the consistency of climate–loess magnetism relationships.

The Duanjiapo section is located at 34.2°N, 109.2°E, in Lantian County on the humid southern edge of the loess plateau (Fig. 1). The present mean annual precipitation is ∼615 mm and the mean annual temperature is ∼13 °C. The loess–palaeosol sequence is about 130 m thick and is underlain by about 30 m of the Pliocene Red Clay Formation. The palaeosols in this and other loess sections in Lantian County are very well developed with strong reddish-brown colouration; they are almost entirely leached of carbonate and often contain reprecipitated nodular carbonate at their base-reflecting exposure to consistently more intense weathering and soil formation compared to sites further north. Previously, the results of mineral magnetic measurements have only been reported for specific intervals of the Duanjiapo section—for example, S0 through uppermost L2 (spanning the last glacial/interglacial cycle) (Florindo et al., 1999) and the L8–S8 couplet (oxygen isotope stages, OIS 20–21) (Guo et al., 2001). The Luochuan section is located at 35.4°, 108.3° E in Luochuan County in the central part of the loess plateau. The current mean annual precipitation is ∼570 mm and the mean annual temperature is ∼9 °C. The loess–palaeosol sequence is ∼135 m thick. It was one of the first loess plateau sites to be studied in detail and has been the subject of detailed stratigraphic studies using a variety of climate proxies; however, previously only magnetic susceptibility measurements have been reported for the entire section.