Journal of Paleolimnology

, Volume 50, Issue 4, pp 561–575

Modification of climate signals by human activities recorded in varved sediments (AD 1608–1942) of Lake Holzmaar (Germany)

Authors

    • Helmholtz-Zentrum PotsdamGFZ Deutsches GeoForschungsZentrum
  • Heinz Vos
    • Institute of Energy and Climate Research: StratosphereForschungszentum Jülich
  • Peter Dulski
    • Helmholtz-Zentrum PotsdamGFZ Deutsches GeoForschungsZentrum
  • Andreas Lücke
    • Institute of Bio- and Geosciences, IBG-3: AgrosphereForschungszentum Jülich
  • Robert Moschen
    • Institute of Bio- and Geosciences, IBG-3: AgrosphereForschungszentum Jülich
  • Norbert R. Nowaczyk
    • Helmholtz-Zentrum PotsdamGFZ Deutsches GeoForschungsZentrum
  • Markus J. Schwab
    • Helmholtz-Zentrum PotsdamGFZ Deutsches GeoForschungsZentrum
Original Paper

DOI: 10.1007/s10933-013-9749-z

Cite this article as:
Kienel, U., Vos, H., Dulski, P. et al. J Paleolimnol (2013) 50: 561. doi:10.1007/s10933-013-9749-z

Abstract

Paleolimnological data from varved sediments in Lake Holzmaar (Eifel, Germany) were combined with documentary data on human activities, long-term data from the Historical Climate Database (HISKLID) for Germany and with recent monitoring data to evaluate changes in deposition that arose from climatic and human influences. The sediment data included seasonal layer thickness in an established varve chronology (1608–1942 AD), subannual chemical element counts, and multiannual organic matter data (TOC, TN, δ13Corg), all combined on an annual scale. Indicators for detritus deposition (lithogenic element counts and detritus layers) determined the first principal component (PC1) of the sediment data. This detritus PC1 was compared to the first PCs of the seasonal precipitation and temperature from HISKLID. While no relation was found to precipitation, the correlation with the temperature PC1 determined by spring to fall temperatures was significant. From 1608 to 1870, a positive correlation of the PCs suggests an increase of detritus deposition in the lake center with increasing non-winter temperatures. These may be linked by lake-internal sediment redeposition that increases when the periods of winter stratification become shorter and that of lake circulation longer. The detritus deposition is modulated by external detritus input depending on the intensity of erosion-conducive land use (wood pasture, wood cutting, and rotational slash-and-burn cultivation). Detritus input diminished when land use slowed down with population decrease as the consequence of plague epidemics, warfare and emigration. After 1870, forest regeneration and improving agricultural practices led to a stabilization of the catchment. Erosion and detritus deposition decreased progressively. The negative correlation of detritus deposition with the gradually increasing temperature presumably mimics a cause-effect relation, although a link with decreasing freeze–thaw action is possible. The modernization of agriculture proceeded with manuring and fertilizing, which caused an increase of lake productivity as indicated by summer blooms of diatoms with enhanced nutrient demand, increased δ13Corg values and sulfur concentrations. Within this well established data base we found combinations of criteria that may be used to deduce natural climatic or anthropogenic influences. The quantitative attribution of these influences remains a challenging task in paleolimnology because their interaction makes the detection of linking mechanisms difficult even at high degree of detail and the processes themselves remain debatable.

Keywords

Varve chronologyμXRF scanningSediment chemistryDiatomsLittle Ice AgeHuman impactDocumentary temperature

Introduction

Lake records from regions that had high population densities in recent centuries have been influenced both by climate and by human activities. Because long-term monitoring data are unavailable for most lakes, nature and timing of lake responses to external factors are difficult to determine (Smol 2010). Climatic and human influences are difficult to disentangle (Fritz 2008), but their distinction can be successful if a number of preconditions are met. It is useful to study multiple sediment variables that track environmental change and to compare the results with data from monitoring studies. An accurate chronology of the sediment sequence is crucial for the comparison of the sediment data with documentary data of local human activities and with regional climate records (meteorological, documentary, or climate proxies) to define a climate baseline. Much of these preconditions are met for the sediment record chosen for this study.

Varve chronology provides a seasonal resolution (internal counting error of 3 %) for the sediment sequence from Lake Holzmaar for the period AD 1608–1942 (Kienel et al. 2005). The sediment variables comprise the varve microfacies (Kienel et al. 2005) focused on the thickness of detritus and diatom-bloom sublayers, subannual element counts (Al, Ca, Fe, K, Mn, S, Si, and Ti) and multiannual geochemistry of organic matter (TOC, TN, and δ13Corg). Statistical analyses to elucidate the combined responses of our data series require equidistant data that are achieved here by transformation of all data to the annual scale.

Samples from soils and bedrock from the catchment were analyzed to track detritus sources. A monitoring study (1995–2004) of meteorological and lake-water variables and deposition in sediment traps (Moschen et al. 2009) provided the data to examine modern inter-relationships.

A number of proxy relations were established and widely used to track environmental change. For example, concentrations of lithogenic elements (Ti, K) and detritus-layer thickness in varves were related to detritus input and rainfall (Enters et al. 2010; Kienel et al. 2009; Lotter and Birks 1997). The thickness of proglacial varves was reported to be related to the summer temperature (Blass et al. 2007; Itkonen and Salonen 1994; Moore et al. 2001). In Lake Holzmaar, lacustrine productivity was found to be reflected by diatom bloom development and assemblage and the δ13Corg of particulate organic matter (POM) (Moschen et al. 2009; Raubitschek et al. 1999).

The advantage of the study period since AD 1600 is the availability of documentary information about Central European climate aside from instrumental coverage of the youngest section. The seasonal precipitation and temperatures in the Historical Climate Database (HISKLID) for Germany (Glaser 2001) serve as climate records for comparison. They show a pronounced period of below-average precipitation in Germany from 1770 to 1820. Pronounced temperature decreases occurred during the Little Ice Age (LIA) and are related to minima of solar activity, the Maunder Minimum (1645–1715) and the Dalton Minimum (1795–1825) of sunspots (Eddy 1976; Hoyt and Schatten 1997; Lean and Rind 1999). The Maunder Minimum delineates the coldest phase of the LIA in Europe with a NW–SE propagation of the cooling by 5–6 °C over Northern and Eastern Europe in winter (Glaser 2001; Manley 1974; Pfister 1999). Moderate temperature decrease was reported for the Dalton Minimum (Koslowski and Glaser 1999; Luterbacher et al. 2004; Pfister 1999).

A rich history record for the Eifel region and the Holzmaar vicinity provides the basis to explore the sediment record with regard to human interaction. Of primary interest is the development of land and forest use (von Haaren 1992; Schwind 1983; Wenzel 1962) in relation with regional economy and infrastructure (Blum 1925; Hesse and Schmitt-Kölzer 1999; Imle 1909) and demography (Graafen 1961; Hesse and Schmitt-Kölzer 1999) because catchment disturbances are recorded in lake sediments in most cases (Dearing et al. 2006; Foster et al. 2003; Zolitschka 1998).

With this well-established data base we examine the timing and nature of responses in the sediment record of Lake Holzmaar and attempt to distinguish natural climatic from anthropogenic influences.

Regional setting

Lake Holzmaar in the West Eifel Volcanic Field (Fig. 1) fills the bottom of the maar crater of a phreatomagmatic eruption dated to 40–70 kyr BP (Büchel 1993). The steepest slopes of the crater are in the north. Tuff eruptions during the maar formation were low. Scoria or pyroclastics were not found during field work within the 2 km2 catchment of the lake. The bedrocks are Lower Devonian greywackes, siltstones and claystones (Meyer 1994) covered by clayey brown soils and planosols (Schwind 1983). Today, two thirds of the catchment are arable land, the remaining area is covered by forest. The crater walls are forested with stands of beech (Fagus sylvatica L.), which date from ca. 1860 in the N, from 1830 in the E, and from 1948 in the S. A small stand of spruce (Picea abies L.) in the S dates from 1932 (Fig. 1, data according to internal documents of the local forestry administration).
https://static-content.springer.com/image/art%3A10.1007%2Fs10933-013-9749-z/MediaObjects/10933_2013_9749_Fig1_HTML.gif
Fig. 1

Lake Holzmaar. a Location in Germany (insert), b catchment situation, catchment area (dashed line) including positions of catchment samples, c bathymetry, forest stands at the lake margin, and position of the cores HZM 41/42

The maximum lake depth is 19 m. A brook, up to 1.7-m deep and almost vertically incised, discharges into a shallow-water area in the SW and leaves the lake as overflow of a dam. The lake is dimictic and meso- to eutrophic (Moschen et al. 2009). During thermal stratification (May to October) epilimnetic pH increases up to 11 accompanied by oxygen supersaturation, whereas anoxia develops in the hypolimnion. An on-lake weather station (installed in 1995) and the meteorological stations Manderscheid and Trier report annual mean temperatures of 8.4 °C, and monthly means of 17.0 °C for July, and 0.3 °C for January. The mean annual precipitation during the monitoring period amounts to 750 mm with maxima exceeding 70 mm in June, August, and December.

Materials and methods

Coring, sediments, chronology, and sampling

Two sediment cores were taken only 3 m apart in the lake center in summer 1999 (Fig. 1). With a total length of 1.07 m, the surface core HZM 41 was recovered with an intact sediment surface using a gravity corer. The core HZM 42 (processed section 0–1.43 m of a total length of 2.39 m) was retrieved using a UWITEC Piston corer (http://www.uwitec.au).

The sediment is macroscopically laminated with light (beige to olive) and dark brown colors. Overlapping thin sections from epoxy-resin impregnated sediment blocks (100 × 20 × 20 mm; ESM Table 1) were inspected by light-microscope and a varve chronology was established (Kienel et al. 2005). The varves are characterized by alternating layers of detritus and of diatoms. Clay to silt-sized clastics are mixed with changing portions of fine organic detritus and periphytic diatoms in the detritus layers and successions of planktonic diatom species form the diatom layers (Kienel et al. 2005). The varves are 0.4–7.1 mm thick (mean: 2.6 mm). Intercalated are 17 turbidites, which are 0.5–14.05 mm thick, grade from fine sand to clay at the top, and contain variable portions of organic detritus. In the top 23 cm of the sediment record, sediment packages with intact varves in a matrix of mixed sediment hint at disturbance of previously existent varves (Zolitschka 1996; Zolitschka et al. 2000). 210Pb dating and varve counting assigned the varved sequence to the period AD 1608–1942 (Kienel et al. 2005).

The cores HZM 41 and HZM 42 are correlated applying two approaches. Visual correlation of prominent layers (turbidites) was done throughout the sequence, while varve-by-varve correlation was confined to the transitions between the cores (Table S1). Within magnetic susceptibility logs measured in 1 mm steps by a Bartington MS2E spot-reading sensor mounted to an automatic logging system designed at the GFZ Potsdam, 44 prominent correlation levels were identified utilizing the interactive core correlation program xtc (Nowaczyk et al. 2012; Electronic Supporting Material ESM Fig. 1). The chronology for the top part of HZM 42 was then established through linear interpolation between these layers.

As sources for detritus input to Lake Holzmaar we sampled topsoils (A-horizon), subsoils (B-horizon) from the brook bank, and bedrock (greywacke and claystone) in the catchment (Fig. 1).

Analytical methods

Subject to element chemistry analyses were the resin-embedded sediment blocks from which the thin sections for varve microfacies analysis were produced. They were scanned with a Micro X-ray fluorescence (μXRF) spectrometer (EAGLE III XL), equipped with a Rhodium tube, at 40 kV and 250 μA. The step size was 500 μm with a 650-μm spot. One point measurement lasted 60 s, including calibration procedures and counting. The age for each measuring point is linearly interpolated within the varve boundaries.

The taxonomy of the diatoms in the bloom layers follows Krammer and Lange-Bertalot (1991) and Håkansson (2002).

Samples for the analysis of sediment dry weight, total organic carbon (TOC), δ13Corg of organic matter and total nitrogen (TN) were taken from core HZM 42. The samples comprise 0.5-cm intervals of sediment of one core half (tube diameter 5.94 cm) for the analyses of TOC and δ13Corg, while TN was analyzed in 1-cm spacing until 87.5 cm depth and in 2-cm spacing below. The material was freeze-dried, homogenized and carbonate was removed for analyses of TOC and δ13Corg. These were performed with an elemental analyzer (Carlo Erba) interfaced in continuous flow mode to an isotope ratio mass spectrometer (IRMS) (Optima, Fisons). Samples weighed in tin foil boats to a total carbon amount of 150 μg were burned at 1,080 °C in excess of oxygen and flushed into the IRMS with helium as carrier gas. Results were calibrated using certified elemental standards and international isotope standards (IAEA, Vienna). Carbon isotope values are reported in the δ notation relative to VPDB in ‰. Analyses of nitrogen content were performed with an elemental analyzer (EA 3000, Euro Vector). The precision for replicate analyses of samples is ±0.2 % for element contents, and ±0.1 ‰ for δ13Corg.

The catchment samples (soils and bedrocks) were dried at 105 °C and pulverized (<63 μm) and followed the above procedure for analyses of TOC, TN and δ13Corg. For X-ray fluorescence (XRF) analysis of major element oxides, the samples were prepared as fused disks of Li tetraborate-metaborate (FLUXANA FX-X65, sample-to-flux ratio 1:6). A Panalytical Axios Advanced wavelength-dispersive spectrometer and matrix correction programs were used to calculate concentrations. Loss on ignition (LOI1000) was determined using a Vario EL III applying high-temperature catalytic combustion at 1,000 °C.

Data analyses

The lake sediment data are of different temporal resolution (Fig. 2), while the seasonal sublayers of varves differ in thickness. The μXRF-element counts and the geochemical data (sediment dry weights, TOC, TN and δ13Corg) were measured in their respective core increments and distances, which represent different time steps and time ranges (Fig. 2a). Their time ranges are approximately log-normally distributed (Fig. 2b) with a median of 0.177 years per μXRF-data point corresponding to 5.65 data points per year and 1.19 years per geochemical sample corresponding to 0.84 samples per year.
https://static-content.springer.com/image/art%3A10.1007%2Fs10933-013-9749-z/MediaObjects/10933_2013_9749_Fig2_HTML.gif
Fig. 2

a Time range per sample of geochemical and μXRF data, b relative cumulative frequencies of sample-time ranges in the three different data types (varve sublayers are annual data), c μXRF-element counts as raw counts and as annual count sums exemplified for Mn, d components used for the annual flux calculation exemplified for TOC

To achieve equidistant data required for statistical analyses, the conversion to annual resolution seemed reasonable even if inaccuracies are inherent to any kind of such transformation. The annual counts of an element [Σcounts year−1], calculated as the sum of the counts of the respective element within the varve, are exemplified for manganese in Fig. 2c. For the geochemical samples, the total sediment flux was calculated as the sediment dry weight divided by core area and included time interval [mg cm−2 year−1]. Organic matter (OM) content is estimated as the product of TOC and the factor 1.87, based on the average composition of organic tissue of marine phytoplankton according to Anderson (1995). Fluxes of TOC, OM, and TN are the respective products with total flux (Fig. 2d). The inorganic matter (IM) flux is calculated as the difference of total flux and OM flux.

Ordination, clustering, and correlation analyses were applied to annual data series, respectively the 11-year running means for the comparison with climate data. Because ordination requires normal distribution, all positively skewed data series have been log-transformed.

Detrended correspondence analysis (DCA) with detrending by segments and non-linear rescaling calculated the length of gradient in standard deviation units (SD) in the response variable (samples). Because the gradient length was short (<3 SD), principal components analysis (PCA) was used to derive the principal components (PC) of the variability in the dataset and to calculate a sample score for each (annual) sample for which the linear regression (model) fits best. To achieve comparability of the species scores (our measured variables), they were divided by their SD. In conjugation with scaling focused on inter-species relations, the biplot depicts the correlations of the species with the two PCs of largest variability. The ordinations were performed with the software package CANOCO 4.5 (ter Braak and Smilauer 2002).

Stratigraphically constrained cluster analysis (Grimm 1987), using the (agglomerative) method of incremental sum of squares, was applied to subdivide the sequence and to facilitate description and comparison with external data.

Results

Subject to analyses for the chemical composition of OM and IM were lake sediments and samples of bedrock and soils from the lake margin and the brook bed of the Sammetbach. The statistical distribution parameters of all data obtained for the lake sediment, including the thickness of varve sublayers and turbidites are given in Table 1. Element count-rates (μXRF) above the background level were measured for Al, Si, S, K, Ca, Ti, Mn, and Fe. A subset of the annual series of element-sum counts [Σcounts year−1], the IM flux, the δ13Corg values and the varve sublayer thicknesses is depicted in ESM Fig. 2.
Table 1

Statistical distribution parameters of all data series, units are for fluxes (mg cm−2 year−1) and for elements (Σcounts year−1)

 

TOC (%)

TOC flux

δ13Corg (‰ VPDB)

IM flux

TN (%)

TN flux

Al

Si

S

K

Ca

Ti

Mn

Fe

Detritus layers (mm year−1)

Spring bloom (mm year−1)

Summer bloom (mm year−1)

Fall bloom (mm year−1)

Turbidites (mm)

Maximum

7.9

18.5

−25.7

706.0

0.98

2.3

702.0

6,695.9

363.4

2,509.5

694.0

2,519.5

857.5

35,262.9

6.4

2.0

3.0

4.2

14.1

Mean

3.6

3.6

−27.3

105.9

0.42

0.4

114.6

1,048.1

26.5

424.5

71.9

354.5

137.3

7,058.3

2.0

0.3

0.1

0.3

3.2

Median

3.5

3.0

−27.3

78.3

0.40

0.4

97.1

876.4

11.9

361.6

59.5

296.3

118.1

6,336.3

1.7

0.0

0.0

0.2

2.8

Minimum

1.8

1.2

−28.3

19.0

0.22

0.1

8.7

148.0

1.1

44.4

9.9

38.1

14.6

1,261.0

0.0

0.0

0.0

0.0

0.5

Skewness

0.83

2.47

0.51

2.71

1.00

2.69

2.58

2.94

4.75

2.82

5.19

3.39

2.79

2.35

1.36

1.80

4.80

3.19

 
The range of δ13Corg of sediment OM is with −28.3 to −25.7 ‰ narrower than that of particulate OM in sediment traps (POM) from the centre of the lake, where maximum values of −19.1 ‰ were reached in summer (Moschen et al. 2009; Fig. 3). Similar to lake sediment values are those measured in catchment soils (A and B horizons) (−29.2 to −25.9 ‰). The TOC/TN ratios of most lake sediment samples vary between 7 and 10 i.e. in the upper range typical for lacustrine algae (Meyers 1997) and are similar to those of subsoils (B-horizon), topsoils (A-horizon), and trap POM. Ratios up to 15 were found only in few samples from the mid eighteenth century. These high TOC/TN ratios hint at contributions of OM from organic rich topsoil (uppermost A-horizon), for which Moschen et al. (2009) found values around 20 (Fig. 3).
https://static-content.springer.com/image/art%3A10.1007%2Fs10933-013-9749-z/MediaObjects/10933_2013_9749_Fig3_HTML.gif
Fig. 3

Cross plot of δ13Corg values [‰ VPDB] and the ratio of TOC/TN measured in lake sediment, trap material, and catchment-soil samples of Lake Holzmaar. Data for organic rich topsoil and trap material are from Moschen et al. (2009)

The samples of bedrocks and soils can be well distinguished according to their SiO2 content with 60–73 % in soils and claystone and 80–86 % in greywacke (Table 2). Claystone and soils are enriched in Al2O3, K2O, and TiO2 compared to greywacke. Eventually, the soils show higher percentages of CaO and OM (LOI1000) than the bedrock samples.
Table 2

Major element oxides [wt%] and OM content measured as loss on ignition (LOI1000) [wt%] of catchment soils (topsoil A-horizon and subsoil B-horizon) and bedrock samples (locations of catchment samples are indicated in Fig. 1)

Samples \[wt%]

SiO2

TiO2

Al2O3

Fe2O3

MnO

MgO

CaO

Na2O

K2O

P2O5

LOI

Σ

Topsoil 1

73.3

0.9

10.9

4.3

0.1

0.8

0.3

0.6

2.0

0.1

6.2

99.7

Topsoil 2

61.9

0.9

17.2

4.6

0.1

1.0

0.2

0.3

2.2

0.1

11.2

99.7

Subsoil 3

69.1

0.9

13.7

3.0

0.0

0.9

0.3

0.5

2.1

0.1

9.2

99.7

Subsoil 4

67.6

0.9

13.4

6.1

0.1

1.2

0.4

0.4

2.3

0.1

7.3

99.8

Topsoil 5

62.1

0.9

14.4

6.5

0.1

2.0

0.6

0.5

2.7

0.1

9.6

99.7

Claystone 1

66.0

0.8

16.0

6.2

0.0

2.1

0.1

0.7

3.0

0.1

4.7

99.7

Claystone 2

64.0

0.8

15.8

8.3

0.2

1.7

0.1

0.7

3.3

0.2

4.8

99.8

Greywacke 1

86.2

0.4

5.4

2.9

0.2

0.5

0.3

1.5

0.5

0.1

1.9

99.9

Greywacke 2

80.1

0.5

7.5

5.6

0.0

1.2

0.2

0.4

0.8

0.2

3.3

99.8

The PCA of the lake-sediment data series resulted in 52 and 15 % of explained variance captured by the first and second principal components (PC1 and PC2; Fig. 4 I). A positive correlation with PC1 is indicated for Si, K, and Ti followed by TOC flux, TN flux and detritus-layer thickness. Further, Al and IM flux, and Ca, Mn, and Fe are related to PC1. The correlation with PC2 is positive for S, δ13Corg, and the layer thickness of diatom summer blooms. Intermediate positions with positive relation to PC1 and PC2 are taken by the layer thicknesses of diatom blooms in spring and fall.
https://static-content.springer.com/image/art%3A10.1007%2Fs10933-013-9749-z/MediaObjects/10933_2013_9749_Fig4_HTML.gif
Fig. 4

I Principal components analysis: correlation biplot of the first and second principal axes of the annual data of element chemistry, geochemistry, and varve sublayers (scores centered and standardized by species). IIUpper panel sample scores of the two first principal components (PC1 and PC2) and their 11-year running means. Lower panel zonation obtained using stratigraphically constrained clustering (Grimm 1987). Cut level one determines the three main deposition phases A, B and C. Cut level 2 determines two deposition sections in phase A and three such sections in phase B. For better visualization, the y-axis is interrupted at three points

The change of sign of the PC1 and PC2 sample scores during the study period indicates distinct differences in sediment deposition (Fig. 4 II). Stratigraphically constrained clustering of the sample scores found three main phases of deposition (A, B, and C) with partitions at 1706 and 1870. Phase A is subdivided into two sections and phase B into three sections.

Discussion

Principal components of deposition

The PC1-related elements Fe, K and Ti are clearly associated with claystones and soils of the catchment (Table 2). Also Ca appears to have a lithogenic origin because authigenic carbonates were observed neither in epilimnion trap samples (Moschen et al. 2009) nor in thin sections.

Questions arise regarding the correlation of TN- and TOC-fluxes with the PC1 that is determined by lithogenic and detritus variables. A similar relation has already been observed in the Late Glacial to Holocene sediments of Lake Holzmaar (Lücke et al. 2003). A first hypothesis would be an external source of OM and IM. For this case, a common means to distinguish terrestrial from lacustrine OM would be the TOC/TN ratio. The compilation of the TOC/TN ratios of potential OM sources for the Holzmaar sediment shows distinctive (increased) values only for organic rich uppermost top soils while the TOC/TN ratios of the other soil samples, bedrock, lake sediments, and trap samples (Moschen et al. 2009) are similarly low and can not be distinguished (Fig. 3). Since diatom bloom layers indicate the importance of autochthonous OM, the correlation of OM and IM flux is thought to relate to the mixing of these components by resuspension and redeposition. The strongest resuspension was observed in Lake Holzmaar during spring and fall in relation to temperature-induced circulation and wind-induced turbulence (Raubitschek et al. 1999; Moschen et al. 2009), similar to observations in lakes of comparable dimensions (Davis 1968, 1973). In summary, the combination of variables captured by PC1 and the above arguments justify the use of PC1 scores as indicators of detritus deposition in Lake Holzmaar.

Captured by the PC2 are δ13Corg, the layer thickness of diatom summer blooms, and sulfur counts (Fig. 4 I). A dependency of δ13C of particulate OM on lacustrine primary production was substantiated during monitoring (Moschen et al. 2009). The diatom summer blooms recognized after 1890 in the sediments (ESM Fig. 2) and in modern sediment traps were formed by Fragilaria tenera (W. Smith) Lange-Bertalot 1980 and Fragilaria crotonensis Kitton 1869 (Kienel et al. 2005; Raubitschek et al. 1999). Both species have a high N demand and indicate sufficient nutrient concentrations (also of Si and P) persisting into the summer (Interlandi and Kilham 1999; Sommer et al. 1986). Along with this indication, the layer thickness of diatom blooms can be used as a measure of diatom productivity (Kienel et al. 2005).

The deposition of sulfur in freshwater lakes is also linked to lake productivity and nutrient concentration. Sulfur is supplied with manure and fertilizer which provide nutrients for lacustrine OM production. This in turn supports the development of anoxic bottom-water conditions and the reduction to immobile sulfides (Holmer and Storkholm 2001). Eventually, sulfur may reinforce eutrophication indirectly when taking the position of P in iron compounds with the result of P release (Caraco et al. 1993; Kleeberg 1997). In modern times, airborne sulfur is added from fossil fuel combustion and smelting activities.

The combination of variables related to the PC2 suggests the use of PC2 scores as an indication of productivity in Lake Holzmaar driven by nutrient input.

Climate relation of detritus deposition

The coverage of the study period by data from the weather station Trier (1851–1897 and 1938–1942) is not sufficient to explore relations between climate and sediment deposition at Lake Holzmaar. We therefore use seasonal temperatures and precipitation compiled in the HISKLID database for Germany (Glaser 2001; Glaser and Riemann 2009; www.hisklid.de) justified by the correlations of the HISKLID data series with Trier temperature and precipitation (ESM Table 2). To account for averaging effects related to temporal lags between mobilization, remobilization and final deposition of material in the lake, we used 11-year-running means to analyze the relations of detritus PC1 with the first PCs of HISKLID precipitation and temperature.

Varve thickness has been reported to relate to precipitation in Finnish lakes (Itkonen and Salonen 1994), but no such relation was found for the period 1957–1987 between Trier precipitation and varve thickness of an earlier cored sediment sequence from Lake Holzmaar with intact varves at the top (Zolitschka 1996). Also for the study sequence only 3 % of the variation of detritus PC1 is explained by the HISKLID precipitation PC1 (not shown). One reason certainly is that the HISKLID precipitation data represent seasonal averages for Germany, which can not resolve local precipitation extremes that are relevant for transport processes.

PC1 of the HISKLID temperature is determined strongest by spring temperatures followed by summer and fall temperatures (Fig. 5 V). The correlation of detritus PC1 and temperature PC1 (Fig. 5 II) is positive with r = 0.36 for the entire study period, but it is not stationary (Fig. 5 VI). We find a positive correlation (r = 0.53) during deposition phases A and B, while it is negative and less strong for the shorter phase C (r = −0.68). In order to eliminate autocorrelation effects due to smoothing, the significance of the correlations has been tested using 11-year-binned data series (23 independent data points for section AB and 6 for section C). The positive correlation in section AB is significant (p < 0.01) independent of the position of the 11 year intervals while the significance of the negative correlation in section C depends on the position of the few data points. This suggests an increase of detritus deposition in Lake Holzmaar with increasing non-winter temperatures in phases A and B and an opposite relation in phase C.
https://static-content.springer.com/image/art%3A10.1007%2Fs10933-013-9749-z/MediaObjects/10933_2013_9749_Fig5_HTML.gif
Fig. 5

Deposition in Lake Holzmaar based on annual sediment data in relation to documentary data, I erosion-favoring factors of human activities, II 11-year running means of the first principal components of Holzmaar-detritus deposition and of documentary seasonal temperatures, grey shaded areas indicate periods with PC1 detritus scores larger than PC1 temperature scores while white shaded areas indicate the opposite, III erosion-reducing factors of human activities, asterisk population numbers from Graafen (1961), IV (complementary) sediment variables not included in the PCA: the TOC/TN ratio and layer thicknesses of turbidites and varves consisting of diatom blooms and detrital layers, V PCA correlation biplot of the documentary seasonal temperature in the HISKLID database (Glaser 2001; www.hisklid.de), VI correlation plot of the sample scores of the temperature PC1 and the Holzmaar detritus PC1

The detritus deposition in the lake center is determined by external input and internal redeposition. A positive correlation between detritus input and summer temperature has been explained via melt-water transport in proglacial settings (Moore et al. 2001; Blass et al. 2007). Transferred to the low mountain range of Lake Holzmaar, it would be the spring temperature and the amount of water stored in snow and frozen soil that would determine the release of melt water and the material mobilization especially from brook bed and incised brook banks (Ollesch et al. 2006). Whether this relation is applicable for the case of Lake Holzmaar has not yet been demonstrated.

A temperature relation of erosion is also established via freeze–thaw action which increases river-bank erosion (Lawler 1986; Ollesch et al. 2006). The inverse relation of freeze–thaw frequency and air temperature (Kreyling and Henry 2011) would imply less river-bank erosion and decreased sediment delivery from the catchment with increasing temperature. A coinciding negative correlation of detritus PC1 and temperature PC1 has been found only in phase C. This observation is in line with the inverse relation of varve thickness to winter/spring temperatures reported for the Holzmaar sequence 1954–1987 (Zolitschka 1996). It is, however, contradictive to the positive correlation of detritus PC1 and temperature PC1 found in phases A and B.

Lake-internal processes, which can produce the positive correlation of temperature and detritus deposition during phases A and B are resuspension and redeposition known as sediment focusing (Likens and Davis 1975). As discussed above, the largest deposition in the center of Lake Holzmaar occurred during spring and fall/winter (Moschen et al. 2009). This was related to resuspension caused by the circulation of the water column, during which the impact of strong winds is enhanced. A prolongation of lake circulation phases would consequently increase resuspension and redeposition. This relation was substantiated for Finnish lakes, where the correlation of varve thickness and spring temperatures was linked to increased resuspension during prolonged circulation periods (Itkonen and Salonen 1994).

The observed change in sign of the detritus-temperature correlation at the beginning of phase C (1870) and the episodic deviation of the detritus scores from the temperature scores (Fig. 5 II) indicate additional influences on the deposition of detritus. Especially external detritus input is strongly determined by the catchment exposure to erosion related to human disturbance of vegetation cover and land use (Dearing et al. 2006; Foster et al. 2003; Zolitschka 1998).

Deposition in relation to human activities

Human activities with the potential to increase catchment exposure to erosion at Lake Holzmaar are compiled in Fig. 5 I. The traditionally used ‘Schiffel’ cultivation, a rotational slash-and-burn-method, is based on turf cutting in the heathland (von Haaren 1992). Soil erosion increased because humus destruction impedes water absorption. Disturbance of the forest floor was caused by wood pasture and harvesting of forest litter. Both gained importance at the onset of the eighteenth century (von Haaren 1992). At the same time, wood cutting increased with the demand of the Eifelian iron industry and led to widespread deforestation (Schwind 1983; Wenzel 1962). Population numbers increased rapidly in the nineteenth century and land use intensified. Because of the high erosion potential of the land and forest use practiced, decreases in population numbers (warfare, plague epidemics, emigration) are evident as slow down of erosion, similarly brought about by forest regeneration (erosion reducing factors in Fig. 5 III). In general, surplus of detritus deposition (grey areas in Fig. 5 II), i.e. detritus scores larger than the temperature scores, is found when erosion was favored, while the opposite is observed when erosion was reduced (white areas in Fig. 5 II).

During section A1 (1608–1658), erosion reducing and erosion favoring influences alternated several times. Erosive land use included wood pasture and “Schiffel” cultivation. Land use slowed down following warfare (Thirty Years’ War from 1618 to 1648) and plague epidemics (Eckert 1996; Mayer 1993; Outram 2001; Wißkirchen 1991).

During section A2 (1659–1705), synchronous with the coldest phase of the LIA, detritus input decreased below the level expected from the temperature decrease because land use slowed down following plague epidemics and emigrations (Mertes 1990). Diatom blooms diminished (ESM Fig. 2) and shifted to fall-blooming species (Puncticulata radiosa (Grunow) Håkansson 2002) after 1670 as a response to decreasing temperatures. The timing in the second half of the Maunder Minimum in the Holzmaar record is consistent with the proposed NW–SE propagation of the temperature anomaly (Luterbacher et al. 2001).

A rapid increase of detritus input at the onset of section B1 (1706–1771) is synchronous with intensified land use after recovery from warfare and unfavorable climate conditions. Harvesting of forest litter for stable bedding (Schwind 1983; von Haaren 1992) and expansive wood cutting for the revived Eifelian iron industry (Wenzel 1962; Schwind 1983) add to the erosion caused by slash-and-burn cultivation and wood pasture.

A series of turbidites between 1737 and 1762 (Fig. 3, Fig. 5 IV) is synchronous with documented construction and repeated repair of the lake dam, regulating the water supply for the downstream mill (Wißkirchen 1991; Fig. 1, ESM Table 3, ESM Fig. 3). The related lake level increases introduced topsoil and terrestrial litter from the newly flooded area, as indicated by TOC/TN values up to 15 for these turbidites (Fig. 3, Fig. 5 IV).

Although the erosion-conducive factors remain unchanged during section B2 (1772–1815) detritus input is lower than expected from the temperature data. Dry springs and summers between 1770 and 1820 (Glaser 2001), also reported to have interrupted mill operation (Wißkirchen 1991), explain this discrepancy. Eventually, land use slowed down with the threat from frequent troop movements and recruitments during the Napoleonic Wars.

Land-use activities intensified with a rapid increase of population (Graafen 1961) and supported the strongest detritus input during section B3 (1816–1869) (Fig. 5 I). Decreases of detritus deposition after 1830 and 1860 can be related to the regeneration of beech forests in the Holzmaar catchment (Fig. 1) after the demise of the iron industry (Schwind 1983).

In phase C (1870–1942), human disturbance diminished. The improved animal husbandry provided manure to replace the slash-and-burn cultivation (von Haaren 1992) and land use slowed down when the “Law on free movement” in 1868 initiated labor migration to the industrial Ruhr district (Graafen 1961). As a result, the catchment stabilized, erosion decreased and less detritus was delivered to the lake. This process is reflected as a general decline of the detritus PC1 scores. Along with the progressively increasing temperature PC1 scores this is thought to mimic a cause/effect relation in the negative correlation between temperature PC1 and detritus PC1.

An additional effect of the modernization in agriculture was the increasing supply of nutrients to the lake, indicated as increased PC2 productivity scores (Fig. 6). Blooms of diatoms with elevated demands for nitrogen (higher than for phosphorus) and an increasing sulfur deposition can be linked to the application of dung after 1850 and of artificial fertilizer, which became available after the construction of the Cologne-Trier railway in 1871 (Blum 1925; Graafen 1961). The 13 % of sulfur in the preferably used ‘Superphosphate’ fertilizer may have also contributed to the sediment-sulfur increase (Eriksen and Mortensen 1999). Although our analyses do not permit the distinction of sulfur species, the development of the annual counts and the timing of peak values are synchronous with the development of additional sources (Fig. 6). Besides the contribution of airborne sulfur from fossil fuel combustion (Smith et al. 2011), the contemporarily largest European lead smelter near Mechernich (60 km NNW of Lake Holzmaar), became an important regional source of SO2 from 1852 onwards (Imle 1909). The relation of lake-sediment sulfur to the NW Eifelian lead-sulfide deposits was established by Schettler and Romer (1998).
https://static-content.springer.com/image/art%3A10.1007%2Fs10933-013-9749-z/MediaObjects/10933_2013_9749_Fig6_HTML.gif
Fig. 6

Annual sulfur (S) counts and the scores of the second principal component (PC2) of the Holzmaar deposition related to lake productivity are compared with the galena lead production of the Mechernich smelter (60 km NNW from Lake Holzmaar, Imle 1909) and Western Europe SO2 emissions (Smith et al. 2011), and supplemented by documentary data of human activity in the lake catchment

At the end of phase C, prominent detritus layers are consistent with earthwork-intensive construction in the catchment. They serve as additional chronological anchor points and include the constructions of the railway track Wittlich-Daun 150 m south of the lake between 1907 and 1909 (Hesse and Schmitt-Kölzer 1999), the military field aerodrome Eckfeld in spring 1939, and the motorway in 1941, which delimits the Holzmaar catchment to the West (Fig. 1).

Conclusions

The origin of changes in deposition at Lake Holzmaar can be tracked based on an annual chronology, well documented climatic changes and human activities. The principal components of deposition are detritus resulting from external input and internal redeposition and the productivity-related deposition of diatom blooms and sulfur.

A relation of detritus deposition with the seasonal precipitation data compiled in the HISKLID database Germany has not been found, possibly because transport-relevant, local events are not resolved in the seasonal and regional average data. A significantly positive correlation of detritus deposition and non-winter temperatures (HISKLID temperature PC1, in order of significance spring, summer, and fall) is found until 1870. The suggested increase of detritus deposition in the lake center with increasing non-winter temperatures can be linked to internal redeposition during prolonged lake circulation. External input of detritus was determined by land-use activities and modulated the temperature-related pattern. Surplus of deposition with respect to this pattern is found synchronous with documented erosion-conducive activities in the catchment such as slash-and-burn cultivation, wood cutting, litter harvesting, and wood pasture. Population numbers determined the intensity of these activities and contemporary with warfare, plague epidemics and emigration detritus deposition was lower than expected from the temperature PC1 scores.

The inverse relation of gradually decreasing detritus deposition and progressively increasing non-winter temperatures after 1870 would be in line with diminished material mobilization by less freeze–thaw action. At Lake Holzmaar, however, detritus input decreased because the catchment stabilized with the regeneration of the forest and the replacement of erosion-conducive slash-and-burn farming by fertilized agriculture. This is thought to mimic a negative cause/effect relation with the progressively increasing temperature PC1. The application of manure and later artificial fertilizer in turn delivered nutrients to the lake as indicated by increased blooms of diatoms with high nutrient demand and the increase of δ13Corg values. Sources for the synchronous increase of sulfur were the ‘Superphosphate’ fertilizer applied and airborne sulfur from fossil fuel combustion and galena smelting in the NW Eifel.

The delineated combinations of criteria from this detailed and chronologically robust data base may be used to infer natural climatic or anthropogenic influences in general. The interaction of these influences however, causes manifold, in part delayed reflection in the sediment record and linking mechanisms remain debatable, even at the high degree of detail.

Acknowledgments

This study is a contribution to the Helmholtz TERENO research project and was initially supported by the Deutsche Forschungsgemeinschaft (Ki 621). We are grateful to R. Naumann for conducting XRF analyses, to R. Breitenbach who provided access to the internal documents of the forestry office in Gillenfeld, and to A. Brauer for valuable discussions. Eventually we acknowledge the constructive remarks of two anonymous reviewers.

Supplementary material

10933_2013_9749_MOESM1_ESM.doc (2.5 mb)
Supplementary material 1 (DOC 2,578 kb)

Copyright information

© Springer Science+Business Media Dordrecht 2013