Sub-decadal- to decadal-scale climate cyclicity during the Holsteinian interglacial (MIS 11) evidenced in annually laminated sediments

To unravel the short-term climate variability during Marine Isotope Stage (MIS) 11, which represents a close analogue to the Holocene with regard to orbital boundary conditions, we performed microfacies and time series analyses on a∼3200-yr-long record of annually laminated Holsteinian lake sediments from Dethlingen, northern Germany. These biogenic varves comprise two sub-layers: a light sub-layer, which is controlled by spring/summer diatom blooms, and a dark sub-layer consisting mainly of amorphous organic matter and fragmented diatom frustules deposited during autumn/winter. Time series analyses were performed on the thickness of the light and dark sub-layers. Signals exceeding the 95 % and 99 % confidence levels occur at periods that are near-identical to those known from modern instrumental data and Holocene palaeoclimatic records. Spectral peaks at periods of 90, 25, and 10.5 yr are likely associated with the 88-, 22and 11-yr solar cycles, respectively. This variability is mainly expressed in the light sublayer spectra, suggesting solar influence on the palaeoproductivity of the lake. Significant signals at periods between 3 and 5 yr and at ∼6 yr are strongest expressed in the dark sub-layer spectra and may reflect an influence of the El Ni ñoCorrespondence to: A. Koutsodendris (koutsodendris@em.uni-frankfurt.de) Southern Oscillation (ENSO) and the North Atlantic Oscillation (NAO) during autumn/winter. Our results suggest that solar forcing and ENSO/NAO-like variability influenced central European climate during MIS 11 similarly to the present interglacial, thus demonstrating the comparability of the two interglacial periods at sub-decadal to decadal timescales.


Introduction
An understanding of the mechanisms and effects of natural short-term (i.e.decadal-to sub-decadal-scale) climate variability is essential for providing projections of possible climate change for the near future.Short-term climate changes are linked to shifts in the modes of variability of the climate system (e.g. the southern and northern annular modes; Stenseth et al., 2003); therefore, a better representation of such climate-mode shifts in climate models may improve simulations of abrupt climate changes (Alley et al., 2003).Although the instrumental record is becoming more valuable as it is lengthened, it is still insufficient to cover the full range of climatic behavior.Specifically, instrumental datasets do not reach beyond the past ∼300 yr (Jones and Mann, 2004), which precludes deeper insights into the underlying physical processes and the evolution of decadal-to sub-decadal-scale climate variability on longer (e.g.interglacial) timescales.In this context, high-resolution palaeoclimate records, particularly from past interglacials that, unlike the Holocene, were unaffected by human interference, can make an important contribution towards elucidating natural short-term climate variability and its future evolution during the present interglacial (e.g.Alley et al., 2003;Brauer et al., 2007;Müller and Pross, 2007;Tzedakis et al., 2009).
Marine Isotope Stage (MIS) 11 is considered one of the best analogues for present and future climate based on long-term similarities with regard to orbital climate forcing, i.e. low eccentricity and dampened influence of precession (e.g.Berger and Loutre, 2002;Loutre and Berger, 2003;Ruddiman, 2005).A number of proxy datasets have provided insights into the long-term comparability between MIS 11 and the present interglacial (e.g.McManus et al., 2003;de Abreu et al., 2005;Helmke et al., 2008;Rohling et al., 2010;Tzedakis, 2010), but owing to a lack of data with sufficiently high temporal resolution, the short-term comparability between the two interglacials has remained ambiguous.
In contrast to most marine records from MIS 11, which typically exhibit relatively low sedimentation rates, varved sequences from lake sediments yield the potential to test whether MIS 11 and MIS 1 exhibit comparable decadal to sub-decadal climate variability.The terrestrial analogue to MIS 11 in Central Europe has long been a matter of heated debate (e.g.Sarnthein et al., 1986;de Beaulieu et al., 2001;Geyh and Müller, 2005; see also Koutsodendris et al., 2010, for a discussion).However, based on evidence from long terrestrial and marine vegetation records from the Massif Central (France; Reille et al., 2000) and off Iberia (Desprat et al., 2005), there is now a substantial body of research that indicates a land-sea correlation of MIS 11c with the Holsteinian interglacial (e.g. de Beaulieu et al., 2001;Kukla, 2003;Nitychoruk et al., 2005Nitychoruk et al., , 2006;;Müller and Pross, 2007;Preece et al., 2007).
In light of the above, we here analyze a ∼3200-yr-long Holsteinian varve succession from the Dethlingen palaeo- lake in northern Germany.In particular, we have performed (i) a detailed microfacies analysis to understand the seasondependent sedimentological processes controlling varve deposition, and (ii) time series analyses on the varve sublayers thickness in order to investigate the short-term climate cyclicity during MIS 11 and to compare it with the instrumental data and palaeoclimatic records of the Holocene.

Material and methods
The Dethlingen palaeolake is located in the Lüneburger Heide region within the lowlands of northern Germany (Fig. 1).After the disintegration of the Elsterian (MIS 12) ice sheet, several deep lakes formed in the vicinity of Dethlingen that were subject to the deposition of diatomaceous, partially annually laminated sediments during the following Holsteinian interglacial (e.g.Benda and Brandes, 1974;Ehlers et al., 1984).Based on the spatial extent and thickness of the Holsteinian diatomite, the size of the Dethlingen palaeolake is estimated to be ∼800 m in length and 300-500 m in width (Benda et al., 1984).The deposits cored at Dethlingen (10 • 08.367 E, 52 • 57.780 N, 65 m a.s.l.) that have yielded the material for this study comprise organic-rich, predominantly regularly and finely laminated lake sediments (Koutsodendris et al., 2010).Here we focus on the interval between 27.93 and 33.68 meters below surface (mbs) that comprises annual laminations spanning the mesocratic forest phase of the Holsteinian interglacial in Central Europe (∼411-408 ka BP), including a prominent centennial-scale climate perturbation, the so-called "Older Holsteinian Oscillation" (OHO; Koutsodendris et al., 2010;Koutsodendris, 2011).
Varve counting and sub-layer-thickness measurements were carried out at 100x magnification on thin sections (size: 120 × 35 mm) using a petrographic microscope.Thinsection preparation followed standard techniques comprising freeze-drying, impregnation with Araldite 2020 epoxy resin under vacuum, sawing, and grinding of the sediment (Brauer et al., 1999b;Lotter and Lemcke, 1999).To warrant continuity of observation, successive thin sections with an overlap of 2 cm were analyzed.
Geochemical measurements were undertaken with a micro-X-ray fluorescence (µ-XRF) spectrometer EAGLE III XL (GFZ Potsdam) using a step size of 100 µm for Al, Ca, Fe, K, Mn, S, Si, Sr, and Ti (50 s count time, 40 kV tube voltage and 400 µA tube current).Measurements were carried out on sediment blocks that had been impregnated with Araldite 2020 epoxy resin.
Time series analyses were carried out on the thickness measurements of the light and dark sub-layers.Multi-taper spectral analysis (MTM) was used for spectral estimation (bandwidth parameter p = 5, and 9 tapers) (e.g.Vautard et al., 1992).The MTM represents an optimal method for producing spectral estimates with high frequency resolution for given degrees of freedom, low bias, and a distribution amenable to the location of confidence levels (Mann and Lees, 1996).In addition, wavelet analysis was applied to identify occurrence intervals and related amplitudes of periodic components of the non-stationary sub-layer thickness time series (Torrence and Compo, 1998).

Structure of varves and depositional processes
The finely laminated sediments from Dethlingen comprise biogenic varves consisting of two discrete sub-layers, a light and a dark one.The boundaries between the light and dark sub-layers are marked by conspicuous changes in their composition.In general, the boundaries between the light and dark sub-layers are somewhat less sharp than the boundaries between the dark sub-layers and the light sub-layers of the following varves (Fig. 2a-c).The distinctly differing composition of the individual sub-layers allows unequivocal definition of their boundaries under a petrographic microscope and thus precise measurements of their thicknesses.In the case of less well-preserved sub-layer boundaries (192 varves), the midpoints of the unclear transitional zones between the distinct parts of the sub-layers have been applied for thickness measurements.The uncertainty range for the thickness measurements of well-preserved varves is ∼0.01 mm and up to 0.05 mm for poorly preserved varves.
The composition and thickness of the light sub-layers are predominantly controlled by the annual cycle of diatom blooms, which is dominated by taxa of the genera Stephanodiscus, Ulnaria, and Aulacoseira.In most cases, the light sub-layers are dominated by one of these genera, resulting in an almost monospecific diatomaceous layer.However, a successive deposition of two sub-layers of different genera during the growing season can be also observed.
The light sub-layers often contain organic matter that increases in abundance towards the boundary with the dark sub-layers.Small-sized (<10 µm) pyrite framboids are often present (Fig. 2d) and occasionally few angular-shaped grains, ranging in size from coarse silt to fine sand, are scattered within the light sub-layers (Fig. 2f).
The dark sub-layers are composed predominantly of amorphous organic matter with fragments of diatom frustules.Reworked periphytic diatoms, plant remains, freshwater sponge spicules from the littoral zone, and chrysophycean cysts are common (Figs.2e, 3a-b).The dark sub-layers often contain low concentrations of clay particles, in contrast to the light sub-layers where fine-grained minerogenic particles are almost absent.
The succession and characteristics of the individual varve sub-layers as described above suggest that the diatomaceous light sub-layers were deposited during spring and summer, whereas the organic-detrital dark sub-layers were formed during autumn and winter (e.g.O'Sullivan, 1983;Lotter, 1989;Brauer, 2004).In particular, the water circulation and high nutrient availability in spring and summer promoted diatom blooms that led to the deposition of diatoms frustules www.clim-past.net/7/987/2011/Clim.Past, 7, 987-999, 2011 10 ìm 10 ìm at the lake bottom, forming the light sub-layers.Stratification of the water column in summer led to anoxic bottom lake conditions facilitating the preservation of varves (e.g.O'Sullivan, 1983;Brauer, 2004).The deposition of diatom frustules, organic matter and other material from the littoral zone of the lake suggests the re-establishment of the lake circulation during the deposition of the dark sub-layers.The mixing of the water column can be attributed predominantly to an enhancement of wind and wave activity during autumn and early winter; in addition, the low content of clay particles in the dark sub-layers points to minor runoff from the catchment area into the lake during that time.The sharp boundary between the dark and succeeding light sub-layer suggests a transient break in sediment accumulation, which may be attributed to an ice-cover of the lake during winter.During that time, single wind-transported coarse silt and sand grains were trapped in the ice and deposited within the lake sediments after ice melting in spring.These dropstone-like sand grains additionally confirm the seasonal interpretation of the sub-layers.The above-mentioned characteristics suggest that the Dethlingen palaeolake was dimictic, ice-covered and stratified during parts of the year, and experienced periods of mixing between these two states (e.g.Lewis, 1983).

Varve counting and thickness measurements
In total, 2864 varves were counted between 27.93 and 33.68 mbs.For small-scale core intervals where varve preservation was poor or sediment had been disturbed during coring or laboratory processing, interpolations were performed based on the average thickness of 20 varves deposited directly below and above the respective interval.Based on these procedures, the floating chronology for the laminated diatomite at Dethlingen was calculated to comprise 3255 varve years.
The average varve thickness is 1.74 mm (Fig. 4).The thickness of the light sub-layers varies between 0.05 and 5 mm (average: 0.68 mm), whereas the thickness of the dark sub-layers varies between 0. (Fig. 4).A qualitative distribution of different types of light sub-layers in the examined core interval was established based on the dominant diatom genera observed in the thin sections; Type-A is dominated by diatoms of the genera Stephanodiscus with a size of >10 µm (Fig. 3c, d), type-B is dominated by elongated diatoms of the genus Ulnaria (Fig. 3e, f), and type-C mainly comprises smallsized diatoms (<10 µm) of the genera Aulacoseira and Stephanodiscus (Fig. 3g, h).In general, representatives of type-A are thicker (average: 0.87 mm) than those of type-C (0.54 mm) and type-B (0.52 mm) (Fig. 4).The distribution of these light sub-layer types within the studied core interval documents a clear succession in diatom assemblages (Fig. 4).

Time series analyses
The power spectra of the datasets for the light and dark sub-layers exhibit several peaks that exceed the 95 % and 99 % confidence levels (Fig. 5).Significant peaks occur at decadal-scale periods of 90, 25, 15, and 10.5 yr, but also at sub-decadal-scale periods of 5.8-6.1,3-5, and 2-3 yr.In addition, the wavelet spectra show a prominent cycle at ∼512 yr for both the light and dark sub-layers (Fig. 6).In the following, we compare these signals with solar cycles and spatio-temporal modes of global climate variability, such as the El Niño-Southern Oscillation (ENSO), the North Atlantic Oscillation (NAO) and the Quasi-Biennial Oscillation (QBO), which are well known from analyses of modern instrumental climate data and the Holocene palaeoclimatic record (e.g.Stuiver and Braziunas, 1993;Mann and Park, 1994;Hoyt and Schatten, 1997;Wanner et al., 2001;Labitzke, 2005;Gray et al., 2010).
Because the solar-like cycles are evidenced in both light and dark sub-layer spectra, we argue that solar forcing influenced the lake sedimentation throughout the year.The light sub-layers at Dethlingen, which represent the primary lake productivity (see Sect. 3.1), are characterized by peaks of all three decadal-scale solar cycles (i.e.Gleissberg, Hale, and Schwabe cycles) at the 99 % confidence level (Fig. 5).This suggests a significant solar influence on the biological productivity of the lake, most likely by affecting water mixing intensity, temperature, light and UV radiation that exert a strong control on algal productivity (e.g.Bothwell et al., 1994;Beer et al., 2000;Graham and Wilcox, 2000;Gray et al., 2010).The occurrence of solar-like cyclicity in the dark sub-layers, particularly the Gleissberg cycle (Fig. 5), points to solar influence on lake circulation during autumn and winter, most likely through atmospheric circulation changes that modulated wind and wave activity (see Sect. 3.1).Possible links between solar irradiance and atmospheric circulation have been attributed to the solar influence on stratospheric temperature that may modify zonal winds and storm tracks (e.g.Haigh, 1996;Carslaw et al., 2002;Gray et al., 2010).
Summarizing the above, the time series analysis of the Dethlingen varve record suggests a strong impact of solar cyclicity on the processes responsible for the seasonal sedimentation by influencing the lake's primary productivity and the atmospheric circulation over the study area.

Variability within the ENSO/NAO band
The Dethlingen varve record reveals significant variability at sub-decadal time scales, with signals exceeding the 95 % or 99 % confidence levels grouped into three distinct bands, i.e. 2-2.7 yr, 3-5 yr, and 5.8-6.1 yr (Fig. 5; Table 1).Most of the significant peaks are recorded in the range of 3 to 5 yr within the conventional ENSO bandwidth (Mann and Park, 1994;D' Arrigo et al., 2005).Variability within the ENSO bandwidth has been reported in lateglacial to recent varve sequences from North and South America (Rittenour et al., 2000;Nederbragt and Thurow, 2005;Fagel et al., 2008), but to date has not been clearly witnessed in varves from Europe.The ENSO is a natural mode of oscillation that results from unstable interactions between the tropical Pacific Ocean and the atmosphere, affecting weather and climate worldwide (e.g.Fedorov and Philander, 2000).A teleconnection between the Pacific region and Europe via the stratosphere allows the ENSO to influence European climate in late winter and spring (e.g.Brönnimann, 2007;Brönnimann et al., 2007;Ineson and Scaife, 2009).The signal in European climate comprises two modes: during El Niño conditions, when a reduction of coastal upwelling and an increase in sea-surface temperature along the western coast of tropical South America is observed in the equatorial Pacific, the European continent witnesses very low temperatures in NE Europe, increased precipitation in the northern Mediterranean region, and decreased precipitation in Norway.Reversed conditions are observed in Europe during La Niña conditions, which comprise the opposite mode of El Niño in the equatorial Pacific (e.g.Brönnimann, 2007).The cyclicity observed at Dethlingen is in agreement with modern weather observations from Europe that suggest an ENSO influence on climate every 3.5 yr (Rodó et al., 1997).The ENSOlike variability is stronger expressed in the spectrum of the dark sub-layers, pointing to a pronounced ENSO impact on winter atmospheric circulation during the Holsteinian interglacial (Fig. 5).
The Dethlingen varve time series further shows significant variability at the margins of the ENSO bandwidth between 5.8 and 6.1 and between 2.4 and 2.6 yr (Fig. 5).Although this variability may again represent an ENSO impact on varve formation, modern observational data suggest that these signals are better attributed to the NAO.The NAO, which represents a hemispheric meridional oscillation in atmospheric masses centered near Iceland and the subtropical Atlantic Ocean, affects European climate particularly in boreal winter from December through March (e.g.Hurrell, 1995;Visbeck et al., 2001;Wanner et al., 2001).The NAO is characterized by a positive mode related to warmer and wetter than average conditions in north Europe and colder and drier conditions in the Mediterranean region, and a negative mode with reversed characteristics.The NAO variability occurs at bandwidths of 2.5-3 and 6-10 yr (e.g.Appenzeller et al., 1998;Hurrell and van Loon, 1997;Pozo-Vásquez et al., 2000).Varve time series from central and western Europe have also reported significant peaks at 6.1-6.2 yr during the Holocene (Livingstone and Hajdas, 2001;O'Sullivan et al., 2002), whereas a similar period at 6.6 yr has been recorded on oxygen isotope variations of calcite varves from the southern Alps during MIS 11 (Mangili et al., 2010).It therefore seems that the ∼6 yr signal documented in European varve sequences represents NAO-like variability rather than ENSO-like variability because the latter is generally more pronounced in the 3-5 yr bandwidth (Mann and Park, 1994).Further evidence for a NAO-like variability in the Holsteinian record from Dethlingen is provided by the fact that the ∼6 yr signal is only evident in the spectrum from the dark sub-layers.It therefore reflects sedimentation processes during autumn/winter, which is in good agreement with the seasonal impact of the NAO on European climate as known from the Recent (e.g.Hurrell, 1995;Visbeck et al., 2001;Wanner et al., 2001).The variability between 2 and 2.7 yr may be attributed to either the NAO or the QBO (Mann and Park, 1994).The QBO is one of the most commonly recorded circulation patterns in modern data, comprising a variability of the equatorial stratosphere expressed by an alternation in the downward propagation of easterly and westerly wind regimes (e.g.Baldwin et al., 2001).Although such periodicities commonly occur in varved sequences, these signals should be interpreted with caution because of their proximity to the 2-yr Nyquist frequency of annual sampling (e.g.Weedon, 2003).
Finally, the spectrum for the dark sub-layers exhibits a periodicity of 15 yr exceeding the 95 % confidence level (Fig. 5).Such a periodicity has been previously noticed in Central Europe during the Holocene and MIS 11, although its forcing has remained unclear (Livingstone and Hajdas, 2001;Mangili et al., 2010).We suggest that this 15-yr cycle may be related to the interdecadal ENSO variability at 15-18 yr (Mann and Park, 1994).This interpretation is further corroborated by modern observations from Iberia that demonstrate the existence of an amplified ENSO signal at 14.2 yr, i.e. after every four ENSO events (Rodó et al., 1997).
To summarize, the Dethlingen varve time series indicates significant sub-decadal climate variability in European climate during MIS 11, which may be attributed to ENSOand NAO-like climate modes.The pronounced signals in the spectrum of the dark sub-layers point to a strong influence of the ENSO/NAO-like variability especially on winter climate, possibly through changes in atmospheric circulation that influenced lake mixing and the duration of ice cover.

Variability of varve thickness through time
The sub-decadal-and decadal-scale cyclicity as described in Sects.3.3.1 and 3.3.2 is evidenced in most parts of the Dethlingen record (Fig. 6).However, a close inspection of the wavelet spectra for the light and dark sub-layers reveals distinct intervals where this cyclicity appears only in one of the two spectra or is discontinuous in both spectra.
An example for the first case is a 430-yr-long interval between 1320 and 1750 varve years (Fig. 6).This interval, hereafter named low-variability interval (LVI), is marked by a strong sub-decadal and decadal cyclicity in the dark sub-layer spectrum and a very weak cyclicity in the light sub-layer spectrum.We therefore hypothesize that although the spectrum for the dark sub-layers points to an external cyclical forcing influencing the Dethlingen palaeolake system, changes in the boundary conditions (e.g.nutrients, water level) during spring and summer precluded the recording of this forcing in the light sub-layer spectrum.To test this hypothesis, we take a closer look at the varve microfacies during the LVI.The onset of the LVI coincides with a major change in the spring-blooming diatom assemblages accompanied by a thinning of the light sub-layers (Fig. 4).In particular, the diatoms dominating the light sub-layers change  (OHO).The position of the OHO is defined by a strong and abrupt decline of temperate tree taxa percentages in the Dethlingen core (Koutsodendris et al., 2010).
from Stephanodiscus (>10 µm) to Ulnaria species, the latter requiring a higher Si:P ratio (e.g.Kilham et al., 1986;Kirilova et al., 2008).It appears likely that the compositional change in diatom assemblages during this interval was caused by a weakening of the spring circulation, thereby decreasing the phosphorus transport from the hypolimnion to the photic zone to the benefit of diatoms that require high Si:P ratio to grow.The lake circulation during this interval may have been modified by changes in atmospheric circulation.As a result, the boundary conditions of the Dethlingen palaeolake were seasonally modified, precluding the light sub-layers to record external forcing.The sedimentation processes became susceptible to the recording of external forcing again when the lake system returned to conditions that supported a stronger blooming of Stephanodiscus species (Fig. 4).
A good example of the second case, i.e. when the spectra of both the light and dark sub-layers do not show variability, is the interval between 2564 and 2782 varve years (29.15 to 28.73 mbs; Fig. 7) that coincides with the prominent OHO event (e.g.Müller, 1974;Kukla, 2003;Koutsodendris et al., 2010;Koutsodendris, 2011).Across the OHO, the wavelet spectra of both the light and dark sub-layers do not show any statistically significant indication for the 11yr Schwabe cycle; moreover, there is a strong weakening of the 22-and 88-yr solar cycles (Fig. 7).In addition, the ENSO/NAO-like sub-decadal variability almost ceases with the onset of the OHO and only recovers again after the end of the event (Fig. 7).Following a similar concept as for the LVI, the absence of short-term variability during the OHO may be explained by seasonal changes in the boundary conditions of the Dethlingen palaeolake.In particular, the composition of the light sub-layers shifts from Ulnaria-dominated to Stephanodiscus-dominated sub-layers with the onset of the OHO, suggesting changes in lake mixing and productivity (Fig. 4).However, the observations from thin sections do not support a scenario of seasonal changes in the dark sub-layers.On top of that, µ-XRF data spanning ∼90 yr before the onset and ∼150 yr into the OHO (29.33 to 28.87 mbs) do not yield evidence for significant modifications in the geochemical signal to support changes in the sedimentation processes during autumn/winter (Fig. 8).In particular, the intensities of minerogenic-detrital indicator elements (such as Al, Ca, K, Ti) and the Si/Al ratio remain rather constant, suggesting no significant change in terrestrial input.In addition, the constant Fe/Mn ratio does not support any oxygenation changes at the bottom of the lake.We therefore suggest that the absence of cyclical signals in the varves during the OHO points to a weakening of the external forcing.If correct, this implies that both the solar activity and ENSO/NAO-like variability were strongly weakened during this period.An alternative explanation could be a weak sensitivity of the climate system to the external forcing during the OHO.
It has been suggested that the triggering mechanism of the OHO may be similar to the 8.  (Koutsodendris et al., 2010;Koutsodendris, 2011).
Based on the Dethlingen time series analysis, it also appears possible that this climate oscillation is further related to lower solar irradiation.This may also have modified sub-decadal climate variability, as it has been suggested for prominent climate oscillations of the present interglacial, i.e. the 8.2 ka event (e.g.Muscheler et al., 2004;Rohling and Pälike, 2005) and the Little Ice Age (e.g.Shindell et al., 2001).The two cases of discontinuous short-term climate variability in certain intervals as evidenced in the Dethlingen varve record highlight the need to apply time series analysis to the seasonal sub-layers thickness measurements.When indications of short-term cyclicities are absent in certain intervals of one of the seasonal sub-layer spectra but present in the same intervals of the other seasonal spectra, the processes involved in varve formation were obviously susceptible to record agents of external forcing only under specific boundary conditions.In contrast, when the cyclic signals are absent from all seasonal spectra, the external climate forcing controlling varve formation was either weakened and/or ceased completely or, alternatively, the climate system was not sensitive to the external forcing during these periods.

Conclusions
Microfacies and time series analyses from an annually laminated sedimentary archive of the Holsteinian interglacial (MIS 11) yield a strong signal of natural cyclicity at decadal and sub-decadal time scales.The decadal-scale cyclicity is attributed to solar forcing that may have influenced the sedimentation of the light varve sub-layers (spring/summer) by driving changes in the productivity of the palaeolake.The sub-decadal-scale cyclicity is attributed to ENSO and NAO climate modes, predominantly influencing the dark sub-layer formation (autumn/winter) through changes in atmospheric circulation that affected lake mixing.Our analyses clearly demonstrate that in order to interpret the signals of varve time series analysis and to correlate them with temporal modifications of the external climate forcing, it is essential to (a) understand the sedimentological processes controlling varve formation and to (b) compare the results of individually analyzed seasonal sub-layer-thickness datasets.
The solar-and ENSO/NAO-like natural cyclicity during MIS 11 as recorded in the ∼3200-yr-long varve time series from Dethlingen is closely comparable with the central European climate variability of the present interglacial.This suggests that the short-term climate cyclicity during the two interglacials is controlled by similar forcing.Taking this observation a step further, we suggest that MIS 11, besides the well-established long-term astronomical analogy, may be regarded as a good analogue for the Holocene with regard to short-term (sub-decadal-to decadal-) timescales.As a result, understanding the short-term climate variability during MIS 11 may potentially contribute to simulate future climate evolution of the present interglacial.

Fig. 2 .
Fig. 2. Thin-section and scanning electron microscope photographs of varves from the Dethlingen core: light and dark sublayers (A) under parallel-polarized light and (B) under crosspolarized light; (C) close-up photograph of successive dark and light sub-layers; (D) pyrite framboids; (E) sponge spicule; (F) windtransported fine sand grain.

Fig. 3 .
Fig. 3. Thin-section and scanning electron microscope photographs from different varve sub-layers from Dethlingen: (A, B) dark sublayer; (C, D) light sub-layer type A; (E, F) light sub-layer type B; (G, H) light sub-layer type C.

Fig. 4 .
Fig. 4. Varve thickness measurements of dark sub-layers and different types of light sub-layers.Positions of the Older Holsteinian Oscillation (OHO) and the Low Variability Interval (LVI) are indicated.

Fig. 5 .
Fig. 5.Power spectra of the light and dark sub-layer thickness measurements.The red line indicates the median red noise; the dashed and solid black lines indicate the 95 % and 99 % confidence levels, respectively.El Niño-Southern Oscillation (ENSO), North Atlantic Oscillation (NAO), and Quasi-Biennial Oscillation (QBO) bandwidths are according toMann and Park (1994).

Fig. 6 .
Fig. 6.Wavelet power spectra of the light and dark sub-layer thickness measurements.Wavelet amplitudes are colour coded from red (high power) to blue (low power).Contoured areas exceed the 95 % confidence levels for a red noise background spectrum.Hatched areas indicate the cone of influence where wavelet analysis is affected by edge effects.Dashed lines mark the solar-like periodicities.White cones indicate intervals with no varve-thickness data.Positions of the Older Holsteinian Oscillation (OHO) and the Low Variability Interval (LVI) are indicated (see Sect. 3.4).

Fig. 7 .
Fig. 7. Wavelet power spectra of the light and dark sub-layer thickness measurements spanning the OHO interval.Wavelet amplitudes are colour coded from red (high power) to blue (low power).Contoured areas exceed the 95 % confidence levels for a red noise background spectrum.Vertical dashed lines mark the boundaries of the OHO at 2565 and 2782 varve years.Horizontal dashed lines indicate the solar-like periodicities.White cones indicate intervals with no varve-thickness data.

Fig. 8 .
Fig.8.Geochemical results of µ-XRF (step size: 100 µm) spanning ∼90 yr before the onset and ∼150 yr into the Older Holsteinian Oscillation (OHO).The position of the OHO is defined by a strong and abrupt decline of temperate tree taxa percentages in the Dethlingen core(Koutsodendris et al., 2010).

Table 1 .
Summary of significant spectral peaks of light and dark sub-layer time series of the Dethlingen varves and their possible forcing mechanisms.
Koutsodendris et al.: Sub-decadal-todecadal-scale climate cyclicity during the Holsteinian (MIS 11) circulation leading to a turnover in central European vegetation