Stary Kaydaky section

5.1.1.1 Glaciofluvial sediments – a loess source for the Stary Kaydaky site

In the Ukraine and nearby in Belarus, respectively, glaciofluvial sediments are continuously present since the Early Pleistocene that means over the whole considered time of loess accumulation (Gozhik, 1995). These areas are drained by the Dnieper River, which is close to the Stary Kaydaky site. Thus the river alluvium and the glaciofluvial sediments in the periglacial desert, at the edge of the ice, are considered as the main dust sources. A confirmation for this assumption is given by the discriminant analysis (Fig. 1-2, Table 1-1) and the element enrichment relative to the UCC (Fig. 1-6). Accordingly, the eolian sediments at Stary Kaydaky are characterized by high contents of the elements Si, Zr and Hf compared

to the near Danube loesses, generally reflecting the contents of the minerals quartz and zircon (Reeder et al., 2006). A similar factor group was identified by Batista et al., (2006) on the FOREGS dataset of European stream sediment, floodplain sediment and soil samples, with high factor scores in the area of the Fennoscandinavian ice sheet. The selective enrichment of these elements in glaciofluvial deposits can be best attributed to the removal of less weathering-resistant minerals during the processes of crumbling and leaching in the sub- and proglacial environment (Lis and Pasieczna, 2006). Whether this mineral sorting is also accompanied with grain size sorting has finally to be proved by texture analyses. However, zirconium and silicon in aeolian sediments are found to be preferentially associated with coarser grain size fractions i.e. coarse silt and sand, respectively (Muhs and Bettis, 2000;

Yang et al.,2006). Further evidence for mineral sorting is given by the A-CN-K plot (Fig. 1-4) and a low Al/Ti ratio (Fig. 1-5), indicating selective enrichment of coarser, more feldsparic material over more clayey, aluminous material (Eissmann, 2002; Lindner et al., 2002).

Deposits of the Dnieper stage (MIS 6), when the distance to the glacier was most proximal, are not distinctly the coarsest in the A-CN-K plot. This contradicts the expectations, if sorting would be controlled by the eolian transport distance. Differences in wind strength due to ice topography could be more important than the absolute distance from the ice margin.

Furthermore, we have to regard the possibility that the mineral sorting is already inherited in the glaciofluvial sediments.

The distribution of sandy and sandy loam soils in the Ukraine (Fig. 1-8) reveals sandy deposits, on the one hand locally distributed close to Stary Kaydaky, on the other hand extensively distributed in about 300 km distance. These may denote both, possible source areas and paleowind direction. The local sand deposits related to rivers are most probably blow outs from the river banks during dry and cold periods of the Quaternary, indicating northerly, katabatic winds from the ice sheet.

Fig. 1-8. Distribution of sand and sandy loam soils (yellow) in the Ukraine, sand and sandy soil texture in Moldova and sand dunes in Romania. The map is based on the soil map of the Ukraine, (http://eusoils.jrc.it, 2006 ), the geomorphologic map of Romania (http://eusoils.jrc.it, 2006 b) and the map of surface sediments of Moldova (Akaдemия Hayk Moлдabckoй CCP, 1978) . The locations of loess-paleosol sites, with published magnetic susceptibility records are given: 1. Mircea Voda (Buggle et al., 2009; this study), 2. Stary Kaydaky (Buggle et al., 2009; this study), 3./ 4.Orsoja/Harletz (Avramov et al., 2005), 5. Mostistea (Panaiotu et al., 2001), 6. Koriten (Jordanova and Petersen, 1999), 7. Durankulak (Avramov et al., 2005), 8. Novaya Etuliya (Tsatskin et al., 2001), 9. Primorskoje (Nawrocki et al., 1999), 10./11./12. Varnitsa/Khadzhimus/Tiraspol (Dodonov et al., 2006), 13. Kolkotova Balka (Tsatskin et al., 2001), 14. Roxolany (Tsatskin et al., 1998), 15. Vyazivok (Rousseau et al., 2001). Arrows indicate proposed paleowind directions during cold stages, according to the distribution of the sandy areas with respect to river valleys. For the Western Ukraine and East Romania, where such deposits are absent, paleowind directions are based on the orientation of gredas according to Rozycki (1967).

In the western Ukraine, these winds were deflected by the Carpathian mountains towards ESE, changing to SE and then to southerly directions at the eastern backside of the Carpathians, as can be deduced from the orientation of gredas (Rozycki, 1967). For the large sand area in the north of the Ukraine, it is not possible to distinguish Quaternary sands and possible autochthonous Paleogene sands. The striking similarity in the zirconium and silicon enrichment of the loess south of it and of Northern European glaciofluvial sediments, however gives reason to favor predominant Quaternary origin. Both, the Dnieper River and the northerly katabatic winds from the ice sheet are seen as responsible for the southward transport of glaciofluvially derived material, in the Ukraine.

5.1.1.2 Origin of the glaciofluvial sediments

Having confirmed that glaciofluvial deposits of the Fennoscandinavian ice sheet with characteristic mineral and grain size sorting are the major loess source, the potential origin of this material is to be evaluated. According to petrographic studies (Dorofeev, 1969; Gaigalas, 1978, 1982; all cited in Matoshko, 1995), far transported material provided by the ice sheet originates from the southern part of the Baltic shield i.e. southern Finland and Gulf of Finland. However, there are also contributions from sedimentary rocks of the Russian platform and locally of crystalline rocks of the Ukrainian shield.

We firstly focused to rule out one of the remaining crystalline areas, but the separation of the potential source areas in the A-CN-K plot was not satisfying. Yet, the Ukrainian shield could

be ruled out as a relevant source due to a low Al/Fe ratio. This ratio is believed to be not affected by weathering or sorting effects, since it is constant for the Dnieper as well as the near Danube loess sections and equal to ratio of the average UCC. The Baltic shield composition fits well with Al/Fe ratio of the loess. This would strongly suggest far-transported material as a major component of the glaciofluvial sediments. Unfortunately, the contribution of the mentioned sedimentary rocks could not be directly evaluated, because suitable data are lacking. However, due to the high extend of initial weathering of the material, we propose that in fact preweathered sediments of the Russian platform account for the major proportion of the glaciofluvial-loess source material. This is not necessarily a contradiction to the observed geochemical similarity between the loess in the Dnieper area and the Baltic shield, showing average UCC like composition, since Taylor and McLennan (1985) observed various terrigenous sediments to be good samples for average UCC. See Section 5.2 for more detailed explanations.