Abstract
We have measured pollen, aluminum, and lead abundances in a Czech peat bog at Bozi Dar in order to investigate the environmental impact of Holocene climate changes and mining activities in Central Europe. The pollen record shows a continuous vegetal cover since 13 kyr BP. Aluminum and pollen deposition characterize the Younger Dryas (YD) cold event at 12.3–11.0 kyr BP, within its expected time frame in Northern Europe. Aluminum fluxes reveal four other significant dust events with significant Asian and Saharan sources, as shown with variations in stable 204Pb, 206Pb, and 207Pb isotopes. Recurrent Asian incursions are seen at Bozi Dar during the YD and Mid-Holocene dust events (at 7.1, 6.5, and 5.9 kyr BP). These Asian imprints are also observed by Le Roux and colleagues during the YD in the nearest Holocene-long geochemical peat record from Switzerland (Etang de la Gruère). We do not see at Bozi Dar the distinct western European volcanic inputs identified in the Swiss peat by Shotyk and colleagues. However, isotopic ratios characteristics of Mediterranean airflows are evident at both locations during the Mid- and Late Holocene. The occurrence of these various air mass sources in Switzerland and the Czech Republic substantiates Holocene atmospheric circulation models in Central Europe with dominating zonal airflows during the Early Holocene, followed by a governing meridian dispersion during the Mid- and Late Holocene. Lead inputs at 5.1–4.3 and 1.9 kyr BP cannot be explained by increased dust deposition only. The 1.9 kyr event fits local Bohemian ore isotopic imprints of which isotopic signatures are seen as early as 2.2 kyr BP in our core, hence revealing the oldest ever environmental record of mining-related activities in the Czech Republic. Relatively elevated Pb deposition at 5.1 and 4.3 kyr BP are tentatively attributed to contamination that may be from local sources and/or long-range transport.
Introduction
Continental climate changes and environmental imprints of human activities during the Holocene are generally identified with long-term sediment, ice, or speleothem records. Among those, ice and peat bog cores, when available, provide more direct and specific records of airborne trace elements, and their stratigraphy is less perturbed. The duration and regional/global extension of these natural or anthropogenic perturbations are best characterized with a multi-proxy approach. In particular, the combination of high-resolution stratigraphy and biogeochemical markers improves characterization of abrupt climate changes, whose duration and threshold intensity are marked, versus other parts of the record that indicate more constant conditions (Alley et al., 2003; Clark et al., 2002; Trenberth, 1997; Trenberth et al., 2004). The former includes one of the most intensively chronicled climate crises during the Upper Pleistocene, the Younger Dryas (YD) cold event, which occurred between 13.0 and 11.5 kyr BP and extended throughout continental Europe (e.g. Alley et al., 1993; Berger, 1990; Broecker, 2003; Clark et al., 2001, 2002; Dansgaard et al., 1989; Peteet, 1992, 1995; Severinghaus and Brook, 1999; Wright, 1989; Wunsch, 2006). That crisis was triggered by meltwater discharge from the Laurentide ice sheet drainage, which caused a partial shutdown of the Meridional Overturning Circulation (MOC; Hughen et al., 1998; Lie and Paasche, 2006; Muscheler et al., 2000) that substantially controls the thermal mass balance of the North Atlantic regions (e.g. Bond et al., 1997; Broecker, 2003; Broecker and Denton, 1989; Fairbanks, 1989; Keigwin et al., 1991; Lehman et al., 1991). Vegetation cover, as characterized by relative pollen abundances, allows identification of other continental past climate episodes, with some being much shorter than the YD event, like the abrupt 8.2 kyr extreme cold episode that has generated drier and colder weather conditions for 200–400 years in Europe (e.g. Allen et al., 1999; Genty et al., 2003, 2005; Veski et al., 2004). Pollen also provide indices for early human activities related to deforestation and/or agricultural practices (e.g. Davis et al., 2003; Heiri et al., 2003; Magny and Begeot, 2004; Magny et al., 2003; Prasad et al., 2009; Rosén et al., 2001; Sarmaja-Korjonen and Seppä, 2007; Seppä and Poska, 2004; Seppä et al., 2005, 2007; Snowball et al., 2002; Spurk et al., 2002; Tinner and Lotter, 2001). Still there are uncertainties in the spatial and temporal distribution of these Holocene climate and early anthropogenic activities as recorded with pollen analyses. Here, we combine pollen abundances along with aluminum (Al) concentrations and lead (Pb) isotopic compositions that are efficient proxies for atmospheric dust and aerosol sources, respectively. Aluminum is a major crustal component that provides a measure of atmospheric dust loads, whose variations between cold and warm spikes are indicative of atmospheric conditions in connection with the extension of desert (lack of vegetation cover) or emerged lands caused by sea-level changes during cold events. Variations in ratios of stable Pb isotopes (204Pb, 206Pb, 207Pb, 208Pb; the last three are end members of uranium–thorium decay chains) can be used to discriminate between various natural (e.g. diverse wind-eroded crustal bedrocks and volcanic emissions) and anthropogenic (e.g. different metallogenous mines) sources because of the specific U–Th content of different crustal reservoirs and ores. Isotopic imprints are preserved during smelting and/or the erosion of geological materials, hence providing an accurate tool to identify the geographical origin of airborne deposited dust and anthropogenic imprints associated with mining/smelting activities in continental reservoirs (Abouchami and Zabel, 2003; Alfonso et al., 2001; Baron et al., 2005, 2009; Brännvall et al., 1997, 2001; Chambers et al., 2012; Chow and Patterson, 1962; De Vleeschouweret al., 2007, 2010; Dunlap et al., 1999; Jones et al., 2000; Jouffroy-Bapicot et al., 2007; Kempter and Frenzel, 2000; Kylander et al., 2005, 2010; Le Roux et al., 2004, 2012; Marcantonio et al., 2002; Martinez-Cortizas et al., 2002; Monna et al., 2004; Novak et al., 2003; Renberg et al., 2001, 2002; Shirahata et al., 1980; Shotyk et al., 1998, 2001; Thevenon et al., 2011).
We have measured pollen abundances, Al concentrations, Pb concentrations and corresponding isotopic compositions in a Holocene-long peat bog from the Czech Republic (Central Europe) in order to assess (1) the inland continental signature of ocean-driven climate events and associated atmospheric circulation patterns; (2) the imprints of volcanic episodes possibly deposited in Central Europe from the main known sources in the Faroe Islands, Iceland, and the French Massif Central; and (3) the earliest discernible environmental imprint of mining/smelting activities in the northwestern Czech Republic, which is one of the richest Ag–Pb European mining districts and where mining has been well documented since the 14th century
Two of the main difficulties in this type of study are to ascertain the geogenic provenance of dust from different sources and whether or not the environmental disturbances recorded within the peat bogs originate from local or distant sources. Here, we shall make use of the most discriminating of Pb, 204Pb, along with 206Pb and 207Pb. This approach is often precluded in peat bog investigations because of the difficulty in accurately measuring 204Pb in environmental samples with low Pb concentrations (Doe, 1970; Kylander et al., 2005, 2010). 204Pb, along with 206Pb and 207Pb, has been commonly used to characterize volcanic and crustal imprints, as well as specific Pb ore signatures in Europe (see references in Allan et al., 2013; Baron et al., 2006; Biscaye et al., 1997, 2002; Garçon et al., 2012; Holm et al., 2001; Kylander et al., 2005, 2010; Stos-Gale et al., 1995, 1996; Thirlwall et al., 2004; Wittig et al., 2007).
This long peat core also provides the opportunity to compare our Czech Republic geochemical record with the other nearest Holocene-long Pb and dust record from Etang de la Gruère in the Jura Mountains, Switzerland (Le Roux et al., 2012; Shotyk et al., 1998; Figure 1). We attempt to corroborate the two data sets and possibly extend findings regarding the occurrence of climate episodes along with the geographical source of deposited aerosols in Central Europe, in particular the occurrence of volcanic events, incursions of Mediterranean and Asian air masses, and the early geochemical record of mining and smelting activities.

(a) Localization of the Etang de la Gruère (Shotyk et al., 1998) and (b) the Bozi Dar (this study) peat cores. Various sized and gray circles represent the oldest Holocene peat cores analyzed for lead (Pb) and its stable isotopes in Europe (12–6 kyr BP according to the caption in the figure). This figure has been modified after De Vleeschouwer et al. (2010).
Materials and methods
Study site and sampling
The Bozi Dar peat bog (15°23′N, 12°54′E) is a rain-fed sphagnum-dominated, mountain-top peatland situated in Ore Mountains (Krusné hory) of Bohemia in the northern Czech Republic (Figure 1). It is a large upland moor located on a watershed that covers an area of 105 ha at the elevation of 993–1035 m. It lies in a moderately warm to cold region with a mean annual temperature of 4.1°C and an average annual rainfall of 1250 mm (Dohnal et al., 1965). The bedrock is composed of micashists and paragneisses, with Neogene nephelinite extrusive rocks covering part of the area. The total thickness of the peat is 3.0 m. The vegetation is dominated by Norway spruce (Picea abies), dwarf pine (Pinus mugo spp. rotundata), European beech (Fagus sylvatica), and dwarf birch (Betula nana). Large sphagnum mats are found in open segments of the bog. The most abundant sphagnum species are Sphagnum riparium Angstr., Sphagnum fallax v. Klinggr., and Sphagnum girgensonii Russ. The herbaceous layer includes Carex limosa, Oxycoccus palustris, Vaccinium myrtillus, Sedum villosum, Pedicularis palustris, Drosera rotundifolia, Arnica montana, and Swertia perennis. The peat is sphagnum-dominated down to 143 cm, and then eriophorum-carex-dominated (Brizova, 1993). This peat was mined between the 18th century and the early 1960s. In 1540
A vertical peat profile was sampled in an intact segment of the peat bog outside the area of historical peat mining in July 1992. A 15-cm-thick layer of relatively dry peat was removed from a 3-m-high cut, and discarded. Compact, dark-colored peat was sampled from the cleaned profile in an upward direction using a plastic shovel for Pb determinations. Approximately 10 g of peat was taken at 5-cm intervals (43 samples in all). Since the uppermost 25 cm of the peat profile may have been disturbed, it was not sampled. Peat samples for Pb analysis were stored in zip-lock™ bags. A full 3-m-deep peat profile for pollen analysis was sampled using metal boxes (10 × 10 × 50 cm3; no peat monoliths could be collected from the depth of 0–80 cm because the substrate was too dry). Sub-samples for pollen analysis were collected with a spatula from peat monoliths at the same intervals as those collected for Pb analysis (43 samples).
Analyses
Pollen was extracted from the peat by HF treatment and subsequent acetolysis (Erdtman, 1954; Overbeck, 1958). The resulting pollen mixtures were placed on a smear glass along with glycerin, ethanol, and water. Over 500 arboreal pollen grains and usually a lower number of non-arboreal pollen grains were counted from each peat section. Pollen abundances are shown in Figure 2.

General vegetation-type phases (Firbas, 1949/1952) and main climate periods (A to E) according to the detailed percentage pollen diagram at Bozi Dar.
Radiocarbon dating of seven depth levels (0.90, 0.95, 1.55, 1.75, 2.0, 2.35, 2.6, and 3.0 m below surface) was made at the Gliwice and Hannover Geochronological Laboratories. The data were calibrated for variable initial radiocarbon concentrations using the OxCal v4.1 and Clam programs (Bronk Ramsey, 2009; Reimer et al., 2009). The median calibrated ages are listed in Table 1, with 2σ calibrated BP ages and relative areas for each 14C analyzed sample. Calibrated ages show a linear correlation (R2 = 0.97) with depth (Figure 3) suggesting a fairly continuous deposition with no observable hiatus. This correlation can be used to calculate accumulation rates (ARs) that are necessary for Al and Pb deposition flux calculations. However, we do not have 14C age determination between 100 and 150 cm (Figure 3) to outline accurate AR. Therefore, we established an age model curve on the basis of calibrated 14C dates in order to assess proper ages and AR thorough the length of the core using the CLAM program (under R programming language, version 2.10.0, copyright©2009, The R Foundation for Statistical Computing; Blaauw, 2010).
Calibrated 14C ages for the Bozi Dar peat core.

Linear correlation of calibrated 14C ages versus depth in Bozi Dar peat core.
Modeled ARs vary from 0.013 to 0.033 cm/yr (with the exception of the top, where 14C date determination may be hampered because of removal of a dried section, mining of the peat for the past 300 years, and surface runoff) with a mean AR of 0.020 ± 0.006 cm/yr (n = 40, 2σ). The lower AR are encountered between 100 and 155 cm (0.014 ± 0.001 cm/yr, n = 10, 2σ) but are close to the mean calculated AR with no noticeable hiatus. Numerical ages are presented in Table 2 along with geochemical data for each sampled depth. They are used in the discussion to define the chronology of encountered pollen, dust, and anthropogenic episodes.
Al, Pb concentrations (ppm), and corresponding deposition fluxes (µg/cm2 and ng/cm2/yr for Al and Pb, respectively), Pb isotopic signatures, as a function of depth, calculated ages (see age model in the method section), and ash content (%) in the Bozi Dar peat core.
‘nd’ means ‘not determined’; statistically significant dust episodes are highlighted in bold.
Pb and Al analyses were performed by Inductively Coupled Plasma Atomic Emission Spectrometry (ICP-AES, Jobin Yvon Ultima C, CEREGE) after acid digestion (HCl–HNO3–HF) of peat material using trace metal clean techniques in a HEPA (class 1000) filtered trace metal clean laboratory. Reproducibility and accuracy for our ICP-AES analyses of major and trace elements have been previously reported (Germanique, 1994; Ollivier et al., 2010). Procedure and analytical blanks were below detection limits. Overall uncertainties on Al and Pb concentrations were generally below 5%. Concentrations are shown in Table 2 and Figure 4. Al and Pb deposition fluxes (Table 2) are calculated using ARs from the CLAM age model and a measured mean bulk density of 0.096 g/cm3.

Aluminum (Al, ppm) and Pb/Al ratios (×104) in Bozi Dar peat core.
A fraction of acid digested peat was purified on ×G1 × 8 anionic resin to extract Pb and measure the ratios of the four stable Pb isotopes (204Pb, 206Pb, 207Pb, 208Pb) by Thermo-Ionization Mass Spectrometry (TIMS, Finnigan MAT 262, CEREGE). Analytical mass discrimination was corrected using concurrent and repeated analyses of a National Institute of Standard and Technology (NIST) standard reference material for common lead (SRM981). Standard deviations of the 206Pb/207Pb ratios did not exceed 0.05%. Uncertainties on 206Pb/204Pb and 207Pb/204Pb ratios were generally below 0.2% and 0.3%, respectively. Lead isotope ratios are presented in Table 2.
Reliability of the atmospheric record
We have to ensure that Al and Pb accumulated within our peat derive from atmospheric deposition only (as expected with ombrotrophic peats) and not from lateral advection or upward migration from runoff and bedrock weathering, respectively. Ash analyses were performed to help identify the ombrotrophic and mineral horizons. Acid-insoluble ashes were determined on separate peat fractions that were dry-ashed at 550°C prior to aqua regia digestion (HCl–HNO3), then filtered through Whatman 42 ashless filter paper and heated again at 550°C according to Vile et al. (1995, 2000). The ash content is below 5% down to 250 cm (Table 2 and Figure 5), corroborating the ombrotrophic character of the peat (Tolonen, 1984). Below, in a fen-dominated section, the ash content reaches 26–92% (Table 2 and Figure 5). This increase could be an indication of the contribution of bedrock weathering to the peat that would interfere with atmospherically deposited Al and Pb of which concentrations increase by a factor 10 or more in this bottom section of the core (Table 2 and Figure 4). We tentatively assess a potential mineral contribution within the peat using the ratio of the crustal element Al to that of ashes. We determine a mean Al/Ash (×10) ratio of 0.8 ± 0.6 (n = 35, 2σ) in the ombrotrophic sphagnum-dominated fraction of the peat (Figure 5). In the enriched ash section of the peat (below 250 cm), we calculate a ratio of 1.1 ± 0.3 (n = 7, 2σ; Figure 5). The ratios are not statistically different (p > 0.1, student t-test) and suggest that the potential minerotrophic nature of the bottom peat does not significantly influence trace metal content in this fraction. The Al/Ash ratios are significantly divergent from these means (p < 0.0001, student t-test) within the very surface (Al/Ash = 4.3) and the very bottom (Al/Ash = 0.01) layers of the peat core (Table 2) likely owing to runoff and drainage at the top (Brizova, 1993) and a possible mineralogical contribution from the bedrock at the very base of the core. The bedrock contribution can also be discerned using the variations of the Pb/Al ratio with depth in the peat. Indeed, any significant deviation of the Pb/Al ratios within the bottom layers of the peat may indicate a contribution in addition to the atmosphere. Here, Pb/Al ratios are generally uniform with the exception of a peak (at 80–95 cm, Figure 4) and three outliers (at 125, 135, and 290 cm; Figure 4). The mean Pb/Al ratios below 250 cm (4.8 ± 1.2 104, n = 5, 2σ) are not significantly different (p > 0.05, student t-test) from the mean Pb/Al ratio in the rest of the core (6.4 ± 2.8 104, n = 24, 2σ), suggesting no interference with the underlying bedrock. The three elevated ratios determined in the ombrotrophic section of the core may reflect the occurrence of pollutant Pb. Indeed, higher Pb/Al ratios generally reveal an anthropogenic contribution to Pb accumulated in sediment cores (Angeledis et al., 2011; García-Alix et al., 2013). These potential pollutant Pb imprints are discussed later. The outlier at 290 cm (Figure 4), corroborates ash results regarding a possible bedrock contribution to accumulated trace elements in the peat and is not considered in our discussion of atmospheric delivered imprints.

Ash (%) and Al/ash ratios (×10) in the Bozi Dar peat core. The vertical dotted line shows the 5% ash limit above which the peat may be considered minerotrophic.
These results suggest that the minerotrophic feature of the bottom peat (below 250 cm) does not seem to prevent the record of atmospheric markers with no noticeable contribution from upward migration of bedrock-derived Pb and Al, with the exception of the very bottom layer at 290 cm. Shotyk et al. (2002) and Monna et al. (2004) show that minerotrophic peats also can preserve atmospherically deposited metal records at the millennia time scale. Meanwhile, this minerotrophic feature, along with compaction, contributes to enhance the bulk density of the sampled material below 250 cm, up to a factor 2 according to Weiss et al. (2002). Because no density was measured in this section, we assume that our flux calculations, using a mean density determined in the ombrotrophic section of the core, likely are underestimated below 250 cm and therefore are not used to calculate a mean deposition flux in the core and for comparison with other published deposition during the same period.
Results and discussion
Pollen abundances
In our description of vegetative phases (I–X), we refer to Firbas (1949/1952), whose classification is most commonly used in Central European regions. The oldest segment of the profile (vegetative phases II to mid-IV) exhibits the lowest proportion of arboreal species. The base of the profile (at 290 cm) is associated with 50% of arboreal species, while mere 30% of arboreal species are found at 260 cm. In younger sections of the profile, pollen of arboreal species predominate (around 75% in vegetative phases V to mid-VIII, and around 85% in vegetative phases mid-VIII to IX). Figure 2 shows an abbreviated BD pollen diagram. In the Late Glacial, the most abundant tree species are Pinus, Betula, Salix, and Juniperus, while herbaceous species (Cyperaceae) prevail. Solely in the Late Glacial, pollen of Hippophae is noted. In vegetative phase IV (Preboreal), the diagram records mixed forests, with Pinus and Betula, and an understory consisting of Juniperus and shrub-like Salix. During the Boreal (phase V), maximum relative abundance of Corylus is recorded. Simultaneously, the abundance of herbaceous species drops significantly. During the Atlantic (phases VI and VII), the relative representation of Pinus is at its lowest. Since the Boreal (V), Quercetum mixtum (QM) peaks in the Atlantic (VI–VII). This group of woody species includes Quercus, Ulmus, Tilia, Acer, and Fraxinus. At the same time, Corylus becomes less common. Also, during the Atlantic, Picea first appears. Alnus and Fraxinus build up from the end of the Boreal through the Atlantic. Unlike Alnus, which remains present throughout the upper parts of the profile, Fraxinus disappears at the end of phase VIII. The onset of the Sub-Boreal (VIII) is marked by a larger representation of Fagus, Abies, and Carpinus. These species remain in the area until the Sub-Atlantic (IX). During the Sub-Atlantic, QM abundance decreases, while Pinus once again becomes most common. The profile records a continuous forested landscape until around 2.3 kyr BP.
Climate-related vegetation changes
In the vertical Bozi Dar profile, vegetative phase I (Firbas, 1949/1952) is missing because no pollen is preserved in the non-organic eluvial sediment. The Late Pleistocene is represented by the Allerod (II) and the YD (III). As seen in Figure 2, Late Glacial is characterized by an open landscape, with grasses-dominated vegetation. The advent of pine and birch coincides with the Early-Holocene climatic warming (Preboreal, IV, 11.8–9.6 kyr BP). Hippophae coincide with the continuing presence of an inorganic admixture in the substrate. These species often suggest periodical colder climatic phases throughout the Holocene. The Boreal (V, 9.6–8.1 kyr BP) is characterized by a further temperature increase. Shade-intolerant shrubs, such as Corylus, are present at the margins of Pinus-/Betula-dominated woods. The Atlantic (VI–VII, 8.1–4.0 kyr BP) represents the warmest period of the Holocene. Pines become less common, restricted to extreme sites, such as rock outcrops and wetlands. We note that pollen does not provide a tool to distinguish between pines and dwarf pines. The climatically demanding group of QM trees peaks in the warmest period (Atlantic). At the same time, the abundance of Corylus declines because of its poor shade tolerance and decreasing nutrient availability. Also, in the Atlantic, Pinus becomes more common in upland areas, at elevations of around 1000 m. The Boreal/Atlantic boundary (here, c. 8.0 kyr BP) is commonly characterized by increasing precipitation (see Brizova, 1999; Mentlík et al., 2010) as evidenced by the advent of moisture-demanding Alnus and Fraxinus. Both species populate numerous riparian zones and also the area of the nearby Komorany Lake. The beginning of the Sub-Boreal (at 4.0 kyr BP) is drier and colder than the previous Atlantic period as shown by the high occurrence of Fagus, Abies, and Carpinus that are temperature-sensitive species (QM). Fagus, Abies, and Carpinus also are frequent with soil maturation and may not be directly related to climate changes. Pollen of Picea and Abies, unlike pollen of most other woody species, is capable of long-distance transport. The abundance of these two species is not necessarily related to local abundances.
According to the observed local pollen arrangements and the Firbas vegetative phases, we infer five local climate trends (A to E) that are shown in Figure 2. They comprise the very cold Late Glacial dominated by Cyperaceae grass land (A, 12.5–11.0 kyr BP), the Early-Holocene warming and the onset of Pinus (B, 11.0–9.0 kyr BP), the temperature increase during the Boreal (V) with a sharp increase of the shade non-tolerant Corylus and concurrent decline of Pinus (C, 9.0–8.1 kyr BP), the warmest and wettest period of the Holocene (Atlantic VI–VII) characterized by the onset of moisture-demanding species Alnus and Fraxinus and the occurrence of matured-soil demanding Alnus and Picea (D, 8.1–4.3 kyr BP), and the drier and possibly colder Sub-Boreal with the decline of temperature-sensitive species (E, <4.3 kyr BP). This climate zonation is used when discussing geochemical data. Pollen records from Bozi Dar and Etang de la Gruère in Switzerland show the same general trends (Roos-Barraclough et al., 2004). Meanwhile, the cold Sub-Boreal period ends earlier in Etang de la Gruère (c. 10 kyr BP) as shown by the occurrence of Corylus, Ulmus, and Quercus. Warmer trends (D) and (E) identified in Bozi Dar with the succession of Alnus–Picea and Fagus–Abies (c. 8.1 kyr BP and 4.3 kyr BP, respectively) are comprised within the same trend in Etang de la Gruère starting c. 7.0 kyr BP.
Dust deposition and climate shifts
We calculate a mean Al deposition flux of 2.66 ± 1.35 µg/cm2/yr (n = 33, 2σ; Table 2) that excludes the minerotrophic layer (below 250 cm) as well as the upper layer where Al deposition is above 20 µg/cm2/yr and could be hampered by lateral advection. According to a student t-test, any flux above 3.5 µg/cm2/yr is significantly higher (with p < 0.001, student t-test) than this mean deposition and is considered a dust episode. Consequently, we can identify five individual dust spikes (D1–D5; Table 2 and Figure 6) with deposition fluxes significantly higher than the mean input. The oldest one (D1) is by far the largest (Al fluxes reach 84 µg/cm2/yr; Table 2) and extends from 280 to 255 cm (i.e. 12.3–11.0 kyr BP). It is comprised within the Late Glacial climate zone A and exhibits the characteristics of the expected YD cold event during the Late Glacial stage with enhanced eolian activity (De Boer, 1995; Kaiser et al., 2007). It is in agreement with the YD interval identified by Shotyk et al. (1998) in Etang de la Gruère (12.5–10.5 kyr). The YD episode at Bozi Dar matches other YD occurrences between 12.9 and 11.5 kyr BP from Greenland ice and Central European lakes and peats (Alley et al., 1993; Björck, 2006; Dansgaard et al., 1989; Enters et al., 2010; Le Roux et al., 2012; Peteet, 1995; Wolters, 2002), thus validating our age model. The second Al peak (D2; Table 2 and Figure 6) is less pronounced with a maximum Al deposition of 5.7 µg/cm2/yr and extends from 10.7 to 9.8 kyr BP (250–235 cm). It covers the warming transition climate zone B (Figure 6). The following maxima D3 and D4 show similar maxima (6.2 µg/cm2/yr) at 5.9 and 3.2–2.8 kyr BP, respectively (Table 2). They both occur during the warm wet and dry climate zones D and E (Figure 6) and therefore are less likely to be associated with cold episodes, but rather northward flows from southern regions. The last Al peak (D5) reaches 20.8 µg/cm2/yr (Table 2). It starts at 60 cm and is post-Medieval.

Calculated Pb and Al deposition fluxes, crustal enrichment factors, and 206Pb/207Pb ratios from the Bozi Dar peat core. We identify (1) various dust events (Di, light gray) defined as peaks (several data points) or outliers (single data point) and (2) anthropogenic events (Ai, dark gray) that are shown with pollen-based climate zones (A to E). Single dates correspond to 14C calibrated ages (Table 1 and Figure 3).
We have identified several dust deposition maxima in our core using Al. While D1 can easily be associated with the YD cold episode, there is no obvious well-known climate event that could be invoked to explain the other peaks or outliers. For example, there is no visible imprint of the 8.2 kyr cold event. Contrary to the YD, the 8.2 kyr episode is not as widely evidenced in continental Europe, possibly because of its shorter duration and intensity (Alley et al., 2003; Wunsch, 2006). For instance, it is not observed north of 60°N in Europe (Bakke et al., 2005; Bigler et al., 2002, 2003; Jones et al., 2004; Korhola et al., 2002; Larocque and Hall, 2003, 2004; Paasche et al., 2004; Rosén et al., 2001; Seppä et al., 2002, 2007), likely owing to a seasonal emphasis of the cooling phases during spring and winter that are not well recorded in septentrional regions. An insufficient sampling resolution may also cause this abrupt event not to be seen in the Bozi Dar record. In order to improve and complement the characteristics of these dust episodes, we use Pb and its stable isotopes that are efficient geochemical tracers of aerosol deposition sources.
Dust sources
The 206Pb/207Pb ratios measured within the peat core are above 1.190, except at 190 cm (8.3 kyr BP) and in the top 95 cm that corresponds to aerosols younger than 2.2 kyr BP (Table 2) and may be associated with pollutant Pb as discussed below. 206Pb/207Pb ratios of dust maxima D1–D4 are not statistically different from the mean 206Pb/207Pb ratio below 95 cm (1.197 ± 0.005, n = 33, 2σ; Table 2). This isotopic imprint is within the range of that measured in other Mid-Holocene (MH) peat bogs from Central and Northern Europe (Klaminder et al., 2003; Kylander et al., 2010; Shotyk et al., 1998) and Greenland ice cores (Biscaye et al., 1997; Burton et al., 2007; Rosman et al., 1997). The D5 dust spike is too young (less than 700 years; Table 2) to be accurately defined owing to uncertainties on the age model. On the basis of 206Pb/207Pb ratios alone, none of the dust episodes can be isotopically identified with certainty. The isotope outlier at 8.3 kyr BP could be associated with specific aerosol sources, but shows no Al deposition maximum that could be related to a dust episode associated with the 8.2 kyr event. In order to resolve the determination of isotopic sources, we use 207Pb/204Pb versus 206Pb/204Pb plot types that allow to distinguish between well-known soil sources from various geographical origins, including local bedrock and long-distance sources from Asia (China, Gobi) and the Mediterranean (Sahara; Figure 7). The local Bohemian Pb ores are also shown to identify local metallurgical activities (Figure 7). Dust isotopic imprints are well scattered within the 207Pb/204Pb versus 206Pb/204Pb plot (Figure 7) that may help resolve specific source imprints for the various dust peaks (identified on the basis of Al deposition fluxes).

206Pb/204Pb versus 207Pb/204Pb plot with dust sources from the Sahara (Abouchami and Zabel, 2003; Hamelin et al., 1989, 1997), the Gobi desert and Chinese loess (Biscaye et al., 1997; Jones et al., 2000), and local bedrock and imprints from Bohemian ores (Vanecek et al., 1985; Veron, unpublished data). The mean and standard deviations (sd) for both 206Pb/204Pb and 207Pb/204Pb ratios are shown in the figure for closed gray squares. Samples with higher standard deviations are either highlighted (sd6/4 < 0.05 and sd7/4 < 0.03) or crossed (sd6/4 < 0.07 and sd7/4 < 0.05) gray squares.
The YD episode (D1) displays a clear Asian imprint that is particularly enhanced at the beginning of this event (12.3–11.7 kyr BP), when dust deposition reaches a maximum (Table 2 and Figure 8). These results corroborate a similar Asian imprint as evoked by Nd isotopes in the Etang de la Gruère (Jura Mountains, Switzerland) high-resolution sampling during the same period (Le Roux et al., 2012). These findings suggest that significant Asian-originating dust plumes invaded all of Central Europe during the early YD period. The local bedrock does not seem to contribute much during the YD, likely owing to remaining ice cover during this post-Glacial period. A possible Mediterranean imprint may also be recorded at Bozi Dar at 11.6 kyr BP. This Mediterranean imprint is recorded further at 10.7 kyr BP and 10.4 kyr BP, at the beginning of the following D2 peak (Figure 8). The strongest flux recorded during this event (5.67 µg/cm2/yr at 9.8 kyr BP) may also retain a significant Saharan component given its closeness to this source imprint. Meanwhile, D2 also reflects some mixing with other source imprints, as seen at 10.1 kyr BP (Figure 8). If we assume a single event for D2, then a mixture of the Mediterranean source with local aerosols (as defined by the bedrock imprint) is more likely to explain the distribution of the D2 isotopic signatures than a mixture with the Asian source that would imply significant wind shifts during the D2 episode. This dust event is also observed in the Etang de la Gruère peat core, where it is partly assigned a volcanic origin on the basis of Nd isotopes (Le Roux et al., 2012). No such source is evident in our Czech core. Indeed, all of the potential volcanic isotopic signatures fall far outside of the observed range of Pb isotopes in Bozi Dar with 207Pb/204Pb ratios below 15.55 for the Icelandic ridges (Elliott et al., 1991; Prestvik et al., 2001; Sun et al., 1975; Thirlwall et al., 2004) and the Faroe Island (Gariépy et al., 1983; Holm et al., 2001), and 206Pb/204Pb ratios above 19.2 for the Massif Central alkali basalts (Pilet et al., 2004; Wilson and Downes, 1991; Wittig et al., 2007). The absence of volcanic signature in our core possibly arises from the far-east location of the Bozi Dar sampling site in Central Europe (as compared with the Jura Mountains) that prevents western European volcanic eruptions from reaching our peat core. For the same reason, we do not detect the dust episodes observed in the Swiss core at 8.4 kyr BP (on the basis of Nd isotopes; Le Roux et al., 2012) and the so-called 8.2 kyr event (from 206Pb/207Pb ratios; Shotyk et al., 1998) in the Bozi Dar geochemical record. An insufficient sampling resolution may also cause these sharp deposition episodes not to be detected at Bozi Dar. The MH period (8.6–6.5 kyr BP) is characterized by distinct Mediterranean and Asian dust imprints at 7.9–7.4 kyr BP and at 7.1 and 6.5 kyr BP, respectively (Table 2 and Figure 8). D3 dust event at 5.9 kyr BP can also be assigned an Asian origin (Figure 6). This imprint is defined on the basis of 14 published Pb isotope analyses that are intertwined between the Saharan and the local bedrock on the 207Pb/204Pb versus 206Pb/204Pb plot (Figure 7). Therefore, any data point positioned within the Asian imprint source may also be interpreted as a mixture between the Saharan and the bedrock imprints and should be considered with caution. The D3 event can be related to the climate Bond Event 4 (Bond et al., 2001). This outlier is also found in the Etang de la Gruère peat (Jura Mountains, Switzerland), where it could not be assigned any specific source origin on the basis of Nd isotopes and 206Pb/207Pb ratios alone (Le Roux et al., 2012; Shotyk et al., 1998). The Mediterranean signature is pronounced during the following D4 episode between 3.2 and 2.8 kyr BP. These results suggest frequent Mediterranean dust incursions reaching Central Europe since 7.9 kyr BP at the onset of the warmer climate zones D and E (Figure 6) corroborating similar findings from Etang de la Gruère (Le Roux et al., 2012; Shotyk et al., 1998, 2001). The discrepancy between the geochemical records in Czech Republic and Switzerland for the westerly (marine 8.2 kyr event, volcanism) and easterly (Asian dust) source imprints suggest a less widely and efficient dispersion of zonal airflows in Central Europe, while similar occurrences of Mediterranean incursions in both peat cores suggest a more efficient meridian transport. This assumption is consistent with circulation patterns determined during the Holocene in Northern Europe from reconstructed lake levels, pollen–climate calibration, and General Circulation Models that propose an efficient zonal circulation during the Early-Holocene followed by more intense meridian airflows during the MH and Late Holocene (see Harrison et al., 1991, 1992; Seppä and Poska, 2004; Yu and Harrison, 1995). Most particularly, prevalent summer time anticyclonic conditions would explain the frequent recurrence of meridian warm Mediterranean incursions that reach Northern European regions during the MH (Antonsson et al., 2008; Seppä and Birks, 2001). This past atmospheric circulation contrasts with the present zonal-dominated airflows established in Northern Europe for the past two millennia. The post-Medieval D5 dust imprint likely shows the contribution of local Bohemian ores with Pb deposition fluxes three to four times as high than during the preceding period. This pollutant imprint shall be further discussed below in the section on ‘Anthropogenic imprints’.

206Pb/204Pb versus 207Pb/204Pb source imprint plot with identified dust (YD, Di) and anthropogenic events (Ai) from Bozi Dar. MH represents the Mid-Holocene period.
Calculated Pb deposition fluxes in our peat core show maxima that generally match dust episodes and are statistically different (p < 0.001, student t-test) from the mean Pb deposition flux (1.96 ± 1.08 ng/cm2/yr, n = 23, 2σ, Table 2) calculated below 95 cm (with 206Pb/207Pb > 1.190) and above the minerotrophic layer (at 250 cm). This mean deposition flux is within the range of the other lowest pre-anthropogenic background measured in European peat bogs (Klaminder et al., 2003; Kylander et al., 2005; Shotyk et al., 2001). Most Pb deposition maxima are clearly from crustal origin as shown by the calculated enrichment factors (EFs; Figure 6). EFs are determined as the ratio of Pb/Al in the sample to that of Pb/Al from a crustal reference chosen here as a mean of continental loess, soils, and upper crust sediments (McLennan, 1995; Wedepohl, 1995). Uncertainties of the calculated EFs reflect the concentration disparities of the chosen mean crustal reference ratio in various continental reservoirs (Figure 6). Our EFs are defined according to Shotyk et al. (1998) for peat material, using Al instead of scandium as a crustal marker for our study. When close to unity, one can assume that the sample is not enriched in Pb compared with a natural ‘pre-anthropogenic’ background, and therefore, increased Pb deposition can be explained by increased dust fluxes. Some EFs clearly extend beyond unity for two Pb deposition maxima at 5.1–4.3 kyr BP (A0) and 2.2–1.4 kyr BP (A1; Figure 6), suggesting the deposition of non-crustal Pb from anthropogenic origin. A0 and A1 show EFs that range from 2.5 to 9.4 (Table 2). These enriched layers are discussed according to Pb isotope imprints in order to corroborate and explain the source of the pollutant aerosols deposited in the peat.
Anthropogenic imprints
The 206Pb/207Pb ratios for A1 (1.173–1.182; Table 2) are significantly (p < 0.0001, student t-test) less radiogenic than the mean pre-anthropogenic 206Pb/207Pb ratio measured below 95 cm (1.197 ± 0.005, n = 33, 2σ). However, 206Pb/207Pb ratios may be occasionally misleading owing to low radiogenic sources from local soil or volcanic emissions (Kylander et al., 2005, 2007, 2010; Rosman et al., 2003; Zheng et al., 2007). The use of the 207Pb/204Pb versus 206Pb/204Pb plot clearly links A1 to the local Bohemian ore source (Figure 8). We may therefore infer that A1 shows evidences of local metallurgical activities as early as 2.2 kyr BP. It is the oldest geochemical evidence for such industry in Czech Republic. Pollen records during this episode display a decrease of QM species that can be partly attributed to mining-related activities and associated deforestation. The other oldest strong geochemical features of metallurgical activity recorded from European peat bogs have been found in Spain at 3.9 kyr BP (García-Alix et al., 2013), in Germany at 3.5 kyr BP (Monna et al., 2000), in France at 3.2 kyr BP (Jouffroy-Bapicot et al., 2007; Monna et al., 2004), and in Switzerland and Sweden at 3.0 kyr BP (Klaminder et al., 2003; Shotyk et al., 1998).
It is difficult to ascertain the geographical origin and extent of anthropogenic disturbances recorded in sediments during the Bronze Age in remote continental regions owing to the fact that (1) most of the anthropogenic markers originate from aerosol transported over long distances, hence inducing weak signatures; (2) most cultivated plants are self-pollinating and do not readily emit pollen in the atmosphere, hence remaining invisible in the regional pollen rain; and (3) a disturbance extending at least 100 ha is needed to generate a visible signal in Arborum Pollen (AP) decline. In spite of these difficulties, environmental imprints of early human activity have been seen in continental Europe from the combination of pollen and Pb records in peat bogs and lakes, in Switzerland at 6.1 and 4.0 kyr BP (Thevenon et al., 2011; Weiss et al., 1997), in Germany at 5.5–3.5 kyr BP (Enters et al., 2010; Monna et al., 2000), in Sweden at 5.4 kyr BP (Klaminder et al., 2003), in France at 5.0 kyr BP (Jouffroy-Bapicot et al., 2007; Monna et al., 2004), and in Spain at 3.9 kyr and 3.2 kyr BP (García-Alix et al., 2013; Kylander et al., 2005). Some uncertainties can be resolved, thanks to the use of Pb isotope when local soil and ore imprints are well known (Monna et al., 2000, 2004). The A0 Pb peaks display 206Pb/207Pb ratios (1.194–1.195; Table 2) that are within the range of the pre-anthropogenic background (1.196 ± 0.006), and therefore, no conclusion could be drawn from 206Pb/207Pb ratios alone. When plotted in the 207Pb/204Pb versus 206Pb/204Pb diagram (Figure 8), they clearly indicate Mediterranean (at 5.1 kyr BP) and local bedrock (at 4.3 kyr BP) sources. Because of a significant increase of EF in these samples, we may infer (1) long-range transport of contaminated aerosols from a distant anthropogenic signal in Mediterranean regions and (2) local disturbances associated with local agricultural activity and/or tentative mining activities. The Mediterranean imprint is also encountered in the layer immediately above A0, at 4.0 kyr BP (Table 2) corroborating the frequent Mediterranean airflow incursions during this period. Our pollen record shows no specific trend that could confirm human activities, neither Cerealia-type species nor strong deforestation evidences. Two 206Pb/207Pb ratios (1.187 and 1.190) that are within the lowest range of the mean 206Pb/207Pb imprint below 95 cm (1.197 ± 0.005, n = 33, 2σ) are measured during the MH at 8.3 and 6.8 kyr BP (MH0; Table 2). The 207Pb/204Pb versus 206Pb/204Pb plot indicates that MH0 imprint tends toward the Bohemian ore source (Figure 8) and may be indicative of some contamination. Unfortunately, no Pb concentration, and corresponding fluxes or EFs, are available at these depths, and therefore, no conclusion can be drawn from the MH0 isotopic imprints alone that could signify the contribution of another non-identified geogenic source. It should be noted that the isotopic imprints measured during the post-Medieval dust event D5 (60–30 cm) likely include some Bohemian ore component (Figure 8).
Conclusion
The most remarkable features from the Bozi Dar peat core are (1) the well-known YD event with specific aerosol transport from Asia at 12.3–11.7 kyr BP, that is also observed at 7.1, 6.5, and 5.9 kyr BP; (2) the recurrent imprint of Mediterranean originating aerosols at 7.9–7.4 kyr BP (before aridification of the Sahara estimated c. 6.0–3.5 kyr BP; Jung et al., 2004; Renssen et al., 2006) and at 4.0–2.9 kyr BP; and (3) indisputable local mining activity starting at 2.2 kyr BP, the oldest ever encountered in Central Europe. The YD event and the Mediterranean airflow incursions are recorded during the same periods in the peat core collected by Shotyk et al. (1998) in the Jura Mountains, Switzerland. Asian sources are suggested by Nd isotopes in the Swiss core, but solely during the strong YD dust deposition episode (Le Roux et al., 2012). More frequent Asian-originating events are encountered in Bozi Dar during dust events of lesser importance and the MH, possibly owing to the eastern inland location of our sampling site compared with the Swiss peat. The inland eastward location of the Bozi Dar peat may also explain why we do not see the 8.2 kyr event as well as volcanic incursions that are detected in the Jura Mountains (Le Roux et al., 2012; Shotyk et al., 1998). Our geochemical markers clearly respond to atmospheric circulation patterns determined from pollen, hydrological, and physical models during the Holocene with (1) a dominant zonal circulation during the Early Holocene with the occurrence of Asian dust plumes in both Switzerland and the Czech Republic and (2) a more efficient meridian dispersion of Mediterranean aerosols during the MH and Late Holocene in Central Europe. Further high-resolution geochemical records from Holocene-long peat cores are needed in central and eastern Europe to corroborate these results, in particular the recurrences of Mediterranean and Asian airflow incursions, the occurrence of abrupt episodes (the 8.4 and 8.2 kyr BP events, volcanism), as well as the environmental imprint of mining activities during the Bronze Age.
Footnotes
Acknowledgements
We are very grateful to G Le Roux for his help with the completion of the 14C age model curve. We thank F Oldfield, JK Cochran, AR Flegal, and CE Lambert for useful editorial remarks. Three reviewers provided very constructive comments that helped improve the manuscript. We acknowledge the support of the OSU-Institut Pythéas at Aix Marseille University.
Funding
This work was supported by the European Commission (grant 244118 SoilTrEC) and the Czech Geological Survey (grants 335600 and 323000).
