Within a 5.5 m thick succession of Upper Burdigalian (Karpatian) sediments
in the North Alpine Foreland Basin (NAFB; Austria), dated to CNP-zone NN4, a
high-resolution section was logged continuously. One hundred samples were
taken with a resolution of
Earlier analyses of geochemistry and calcareous nannoplankton assemblages hint at small-scale, short-term variations in paleoenvironmental conditions, such as water-column stratification, primary productivity, organic matter flux, bottom-water oxygenation, freshwater influx, and changes in relative sea level. The results indicate a highly dynamic shallow marine setting that was subject to high-frequency environmental changes on a decadal to centennial scale.
Time-series analyses on nine different proxy data sets using REDFIT analysis and wavelet spectra were applied to resolve a possible cyclic nature of these variations. Analyses revealed that different proxies for precipitation, upwelling intensity, and overall productivity were likely controlled by different cyclicities.
A best-fit adjustment of the likely sedimentation rates within the
high-resolution section resulted in periodicities fitting well with the Lower
(
For the first time, short-term climate variability on a decadal to centennial scale is resolved in Lower Miocene shallow marine laminated sediments in a land-based section. The results hint at a close relationship between climate variability and solar forcing during the Late Burdigalian. Moreover, accepting that these cyclicities are indeed of solar origin, this would indicate that precipitation was driven by the two Gleissberg cycles, while upwelling was driven by the Suess cycle. Furthermore, proxies for primary productivity were influenced by both cycles, although the Suess cycle exerts dominant control, reflecting a stronger influence of upwelling on primary productivity.
Instrumental documentation of high-frequency climate behavior only reaches
back several hundred years (see Le Treut et al., 2007, for an extensive
discussion on the subject). Studying sediments offers a unique way to
reconstruct climate variability in Earth's past, also hinting at the
processes influencing such changes. Astronomical parameters are a major force
of climate change on different scales. On a centennial to decadal scale, solar
variation appears to be one of the key factors in determining the Earth's
climate (Patterson et al., 2004, 2013; Solanki et al., 2004; Gray et al.,
2010). While its causes are, for the moment, only poorly understood, solar
activity clearly shows quasi-cyclic variances such as the 11-year
Schwabe cycle, the 22-year Hale cycle, the 50–80-year Lower
Gleissberg cycle, the 90–140-year Upper Gleissberg cycle, the
While cycles of solar origin are well documented in the Antarctic isotope record and ice cores (e.g., Damon and Sonett, 1991; Stuiver and Braziunas, 1993), terrestrial and marine records around the globe also show evidence of solar cycles during the Quaternary (Stuiver and Braziunas, 1993; Stuiver et al., 1995; Taricco et al., 2009; Di Rita, 2013; Galloway et al., 2013). Comparable studies from the Neogene are scarce and mostly confined to varved (Vos et al., 1997; Lindqvist and Lee, 2009; Lenz et al., 2010) sediment. Another example is laminated sediments of Miocene lakes (Gross et al., 2011; Kern et al., 2012, 2013; Harzhauser et al., 2013). These examples show clear evidence of solar cycles in the Miocene. However, studies of marine sediments on high-resolution, short-term climate variability have never been conducted in pre-Pleistocene sediments.
Generally, only specific proxies are appropriate for this type of study. Besides geochemical and geophysical proxies, only microfossils occur in suitable quantities and are small enough to allow quantitative high-resolution studies on a sub-Milankovitch scale. Based on their small size, abundance, and global distribution, calcareous nannofossils are thus uniquely suited for such studies (e.g., Ziveri et al., 2004).
While calcareous nannofossils are widely used in stratigraphy from the Mesozoic to recent, they are less often used for ecological studies, since their complex ecological preferences, while well documented, are still poorly understood (e.g., Ziveri et al., 2004). However, careful literature studies in combination with statistical analyses open up their use as a reliable proxy for the description of shifts in paleoenvironmental conditions (e.g., Couapel et al., 2007; Auer et al., 2014). Nevertheless, high-resolution studies using calcareous nannoplankton as a paleoecological proxy in the Pleistocene and Holocene have shown its potential as a proxy for decadal to centennial climate variability (Negri and Giunta, 2001; Álvarez et al., 2005; Mertens et al., 2009; Incarbona et al., 2010). It further has been proven that calcareous nannofossils are strongly influenced by cyclic variations in the Earth's climate, both on a larger Milankovich scale (Amore et al., 2012; Girone et al., 2013) and smaller scale linked to solar variation (Incarbona et al., 2010). Moreover, a tight relation to solar activity has to be expected for surface-dwelling photosynthetic organisms.
Additionally, nearly all these studies have been carried out on core samples, while calcareous nannofossils from Miocene surface outcrops have never been studied before at this resolution with a focus on the influence of solar variation on short-term climate variability. One previous study on finely laminated marine sediments of late Burdigalian age from the Central Paratethys has already documented ecological variability at such a high resolution during the late Early Miocene. A section was studied in terms of distribution of calcareous nannofossil assemblages in combination with geochemical proxies. Based on this multi-proxy approach, a clear link was shown between abundance distributions within the coccolith assemblages and changes in local environmental conditions, such as relative sea level, freshwater influx, nutrient availability, and upwelling intensity, despite contamination of certain taxa by reworked specimens (Auer et al., 2014).
This data set has now been used to detect a possible cyclic nature in these changes and to correlate them with known decadal to centennial cycles of solar origin. Our results demonstrate that solar forcing had a strong influence on the local paleoclimate and acted as a major pacemaker for the observed cyclic variations in the paleoecological conditions within the area.
Map showing the position of the studied outcrop near Laa an der Thaya in Lower Austria (after Auer et al., 2014).
The investigated outcrop is situated in a brickyard near Laa an der Thaya,
roughly 50 km north of Vienna, Austria (48
Stratigraphic table of the Early to Middle Miocene, showing global geochronology after Gradstein et al. (2012) and references therein. Regional stages of the Central Paratethys after Piller et al. (2007), with updated ages for the Ottnangian and Karpatian after Grunert et al. (2010b) and Dellmour and Harzhauser (2012). The grey bar indicates the stratigraphic position of the Laa Formation (after Auer et al., 2014).
The Laa Formation corresponds to the lower to middle Karpatian (17.2–16.5 Ma; Dellmour and Harzhauser, 2012) of the regional chronostratigraphic scheme of the Central Paratethys. Grunert et al. (2012, 2010b) suggest a total range of 17.2 to 15.9 Ma for the Karpatian (Fig. 2). Seismic logs indicate a maximum thickness of up to 1000 m for the Laa Formation, which displays a general coarsening-upward trend (Roetzel and Schnabel, 2002; Adamek et al., 2003; Dellmour and Harzhauser, 2012).
The Laa Formation is composed of marine, calcareous, laminated, greenish to brownish grey, micaceous silty clays with thin, fine sand intercalations that unconformably overly Ottnangian sediments (Nehyba and Petrová, 2000; Roetzel and Schnabel, 2002; Adamek et al., 2003).
Paleoenvironmental reconstructions suggest an inner to outer shelf environment of 100 to 200 m water depth, with dysoxic bottom water conditions, based on benthic foraminiferal assemblages (Spezzaferri and Ćorić, 2001). Roetzel and Schnabel (2002) interpreted frequently occurring sandy intercalations as episodic storm events. Assemblages of calcareous nannoplankton point towards cool to temperate, nutrient-rich surface water conditions and local upwelling (Spezzaferri and Ćorić, 2001). Similar upwelling conditions have also been reported from other locations within the NAFB during the Early Miocene (Spezzaferri et al., 2002; Roetzel et al., 2007; Grunert et al., 2010c, 2012).
Nannofossil assemblages place the Laa Formation entirely into Neogene Nannoplankton Zone NN4 of Martini (1971) (Spezzaferri and Ćorić, 2001; Spezzaferri et al., 2002; Adamek et al., 2003; Svabenicka et al., 2003), or alternatively into Mediterranean Neogene Nannoplankton zone MNN4a (Fornaciari and Rio, 1996; Fornaciari et al., 1996) and also Calcareous Nannofossil Miocene Zone CNM6 of Backman et al. (2012) (Auer et al., 2014). The assemblages are generally of low to moderate diversity with high amounts (up to 45 %) of reworked (Paleogene and Cretaceous) specimens (Auer et al., 2014).
The studied outcrop was biostratigraphically dated using calcareous
nannofossils, confirming the biostratigraphic position within MNN4a and CNM6
by the occurrence of
In the clay pit a 5.5 m thick succession of finely laminated blue-grey to green-grey clays with intercalated silt and fine sand layers is exposed. The laminated clays show variable thicknesses on a sub-millimeter to centimeter scale. The sand layers, reaching a thickness of up to 5 cm, decrease in frequency towards the top of the succession, and show well-preserved current ripples. The exposed sediments are mostly free of bioturbation as well as macrofossils.
Besides lithological logging, natural gamma radiation and magnetic susceptibility were measured using a portable scintillation counter (Heger-Breitband-Gammasonde) and a portable magnetic susceptibility meter (Exploranium KT-9), respectively.
Within this succession, a high-resolution sub-section with a total thickness
of 940.5 mm was logged, composed of finely laminated clays with some
intercalations of fine sand and silt. The sub-section starts at 3.06 m above
the base of the section and contains 100 layers with a thickness of
In the lower half of the section the clay is clearly laminated. Lamination becomes less pronounced towards the top, while sandy/silty intercalations increase in frequency. This general coarsening-upward trend is also reflected in the geophysical measurements. Furthermore, the lower boundaries of the layers become less clearly defined and increasingly wavy. Bioturbation only occurs in two successive layers close to the base of the section (Fig. 3).
Lithologs of the 5.5 m sedimentary succession and the high-resolution (HR) section. The position of the high-resolution section within the succession is marked by the grey rectangle; the red line denotes a distinct marker layer that was used as a tie point. Relative sea level was reconstructed using gamma-ray logs. Gamma-ray logs (in CPS) and magnetic susceptibility of the outcrop are shown as black lines. A three-point running mean for both logs was plotted as a mirrored column (after Auer et al., 2014).
For geochemical analysis, approximately 0.1–0.15 g of powdered sediment for each of 100 samples was analyzed in a LECO CS230 analyzer in order to determine
the weight percent of total carbon (TC), total organic carbon (TOC), and sulfur. The
content of total inorganic carbon (TIC) was then calculated based on TC and
TOC content (TIC
For analyses of the calcareous nannofossil assemblages, smear slides of the
samples were prepared after the standard preparation methods outlined in Bown
and Young (1998). Samples were not treated beforehand in order to preserve
the original assemblage and ultrasonicated for 5 s in order to facilitate a
better disarticulation of the sediment before transferring the suspension
onto a cover slip. The slides were mounted using
Eukitt® before being studied under a
standard light microscope with a magnification of 1000
Sedimentological (% of grains
Overview of the nine studied proxies. Table shows highest and lowest values, average content (arithmetic mean), and the standard deviation.
Calcareous nannoplankton assemblages were identified using the Nannotax website (Young et al., 2013) in combination with the taxonomy of Perch-Nielsen (1985b, a), Young (1998), Varol (1998), and Brunette (1998), supplemented by the Handbook of Calcareous Nannoplankton 1–5 (Aubry, 1984; 1988, 1989, 1990, 1999). The revised taxonomy for the Paratethys published by Galović and Young (2012) was also taken into consideration.
Recorded taxa were first identified to species level and then grouped according to their stratigraphic range in order to quantify the amount of allochthonous taxa present within the assemblages, and also to create a stratigraphic framework for the section.
For the genus
A total of 124 calcareous nannofossil taxa were identified, of which 24
occur within nannoplankton zone NN4 and represent an average of 67.86 %
(
The remaining 100 taxa are all clearly allochthonous of both Paleogene and Cretaceous age. Forty-one taxa can be attributed to mainly Paleocene and Eocene ages. The Cretaceous assemblage is represented by 59 taxa of mainly Campanian to Maastrichtian age. All 100 taxa of the allochthonous assemblages were grouped together and used as a proxy for terrigenous input in subsequent analyses (Auer et al., 2014).
Relative abundances of autochthonous coccolith taxa
Since any observed abundance of calcareous nannoplankton is not normally distributed, the counts were transformed using the arcsine transformation, as outlined in Sokal and Rohlf (1995). The reason for using a transformation was to create a more normally distributed data set that is better suited for the subsequent analyses.
Spectral analyses of geochemical and paleobiological data sets were performed using the REDFIT tool of the PAST statistics package (v. 2.17; Hammer et al., 2001). The REDFIT analysis was selected based on the fact that REDFIT was specifically designed to handle unevenly spaced noisy data (Schulz and Mudelsee, 2002). This made the method ideal for the analysis of data sets based on layers of varying thickness.
Prior to analysis, the samples were evaluated based on their sedimentological, geochemical, and taphonomic characteristics and separated into three depositional intervals (Figs. 4, 5). Based on this simple evaluation, the topmost 30 samples of the section were excluded due to intercalations of coarser sediment indicating changes in the sedimentation rate (Figs. 4, 5). Such strong disturbances in sediment accumulation are usually detrimental to the detection of clear signals of periodic cycles in the section (Weedon, 2003). For the remaining 70 samples, with an overall thickness of 684 mm, an on average stable mass budget of sediment deposition was assumed that was not directly affected by episodic storm events or heightened riverine input caused by episodic floods. While sedimentation rates naturally were not fully constant, minor quasi-random changes in sedimentation rate do not adversely affect the methods used (see Weedon, 2003). Within REDFIT, such variations in sedimentation rate are simply expressed as broadened peaks in the power spectrum, indicating that cycles are subject to slight variations in frequency over time. The peaks are, however, still centered on the significant frequency (Weedon, 2003).
In order to consider the thickness of the sampled layers, two data points were used for each sample. One data point was set at the bottom and one at the top of the layer. This was done to express each sample as a continuous layer instead of one single data point. In total, nine data sets (three geochemical and six based on coccolith abundance data) were analyzed using REDFIT. The five autochthonous coccolith taxa used were selected due to their high abundance and specific ecological preferences. The amount of allochthonous taxa was used as a proxy for terrigenous input.
Periodograms of the nine investigated geochemical and paleoecological proxies showing their respective periodicities in millimeters calculated using REDFIT. Black lines indicate the power of the frequency components. Red and orange lines respectively show the 99 and 90 % confidence intervals (CI) of the REDFIT analyses. Grey bars represent the reported periodicities of the Lower and Upper Gleissberg cycle, and the Suess/de Vries cycle.
Overview of the setting used in REDFIT for the nine studied proxies.
Summary of results of the REDFIT analysis for the nine analyzed
proxies. Table shows the detected frequencies, their confidence intervals
(Monte Carlo corrected), and the resulting periodicity in millimeters.
Periodicities in millimeters were then transformed into periodicities in years using a
sedimentation rate of 575 mm kyr
For the calculation of the REDFIT periodograms, different windows as well as segmentations and oversampling rates were applied to all data sets (see Table 2). This was done in order to achieve clear peaks in the periodograms for each separate data set (Hammer, 2010).
After performing the REDFIT analysis on the data sets of all relevant samples,
a Monte Carlo simulation was performed on the resulting power spectra in
order to test the time series under white-noise conditions before examination
of significant peaks (Hammer et al., 2001; Hammer, 2010). Peaks were
considered significant with a confidence interval (CI) of
In order to eliminate any artificial peaks created by the REDFIT analysis,
the Nyquist frequency was used. The Nyquist frequency is the highest possible
frequency in a sampled set that can be expressed by a simple sine or cosine
wave, and is generally assumed to be a frequency of twice the average
sampling distance (Weedon, 2003). Consequently, the Nyquist frequency is
calculated using the reciprocal value of the total thickness of the section
(684 mm) divided by twice the amount of samples used (70). For the studied
high-resolution section the Nyquist frequency can thus be calculated as
1/(684/(70
An automatic linear interpolation was used to create data sets with 1000 new, evenly spaced data points without compromising the shape of the original curve. These data sets were used for methods (filtering and wavelet analyses) that require evenly spaced data sets.
Periodicities, as detected by the REDFIT periodograms, were filtered using Gaussian band-pass filter in the application Analyseries (Paillard et al., 1996). The method selects only the desired frequency components of a given data set, and displays their amplitude, while removing all other frequencies. To encompass the whole range of the selected frequencies, a bandwidth with the size of 25 % of the frequency was applied to all Gaussian band-pass filters. Band-pass filtering is now standard practice for modern cyclostratigraphy, since it allows for a better examination of the varying degrees of influence the detected frequencies have over a time interval (Weedon, 2003).
To further evaluate any detected and filtered periodicities wavelet analyses were also performed on all investigated data sets using the corresponding tool from the PAST statistics package (Hammer, 2010; Hammer et al., 2001). wavelet spectra allow for simultaneous graphic examination of all found frequencies in combination with their amplitude at each data point of a given data set. This method was developed to accurately trace possible frequency and amplitude modulations of detected periodicities throughout a time series (Torrence and Compo, 1998; Hammer et al., 2001; Weedon, 2003; Hammer, 2010). Assuming stable frequencies over time, this further allows for estimations of occurring changes in sedimentation rate.
Geochemical data sets (calcium carbonate, sulfur, and TOC): data were transformed to reflect thicknesses of the sampled layers and filtered according to the periodicities detected by REDFIT. Gaussian filters applied to the periodicities use a bandwidth of 25 % of the filtered frequency. wavelet spectra were used to evaluate the frequencies for spectral shifts; red areas of the wavelet indicate stronger amplitudes.
Following the methods used by Kern et al. (2012), the correlation with known cycles in the centennial to millennial band was achieved by a best-fit adjustment of likely sedimentation rates. The rates were initially estimated from both the calculation of the maximum mean sedimentation rate of the Laa Formation and comparison of sedimentation rates in similar recent sedimentary environments. Based on these initial constraints on sedimentation rates, average rates were adjusted uniformly for the whole section to tune the detected periodicities to known decadal to centennial cycles of the solar band. This approach preserves the relationship between the detected periodicities, which can then be fitted to the characteristic spectral pattern of known cycles (e.g., Weedon, 2003; Kern et al., 2012).
For the detection of possible cyclicities in the high-resolution section, nine
relevant data sets were selected for the REDFIT analysis (Fig. 6). These
data sets include geochemical (calcium carbonate, sulfur, TOC; Table 1) and paleobiological proxies, with the five autochthonous taxa
Calcium carbonate displays significant peaks at periodicities of 91.84 and 153.08 mm, as well as 38.81 and 30.96 mm. The relative proximity of the latter two periodicities indicates that they may represent two different frequency components of the same periodicity, and therefore they were considered further (Fig. 7), even though they do not cross the 90 % confidence level. Their inclusion is also supported by the appearance of similar cycles in related proxies. Filtering indicates that the 91.84 mm periodicity has an amplitude modulation indicative of a very strong influence in the middle and upper part of the section (Fig. 7a). This pattern hints at a steady increase in influence over the section. The amplitude modulation for the 153.08 mm periodicity reveals an increase in the mid-part of the section, where it nearly doubles in intensity, staying constant until the top of the section (Fig. 7a). The two periodicities of lower significance (38.81 and 30.96 mm) show a remarkably similar amplitude modulation over the section, further hinting at their close relationship to each other. Their influence dominates the lower part of the section, which is followed by a strong decline before increasing in amplitude synchronously with the 91.81 mm periodicity, followed by a gradual decline towards the top of the section (Fig. 7a). The wavelet analysis also supports the filtered periodicities detected by REDFIT. Other frequencies not detected by the REDFIT analysis also occur in the wavelet but were thus not considered significant (Fig. 7a).
Sulfur yields two peaks at periodicities of 125.25 and 44.44 mm, respectively. Another less significant peak occurs at a periodicity of 36.26 mm (Fig. 6). Applying a Gaussian filter to the periodicities revealed that the 125.25 mm periodicity dominates the lower part of the section but is much less pronounced in the middle and upper part. The two higher periodicities of 44.44 and 36.26 mm show an inverse pattern. They also show a similar pattern in amplitude modulation over time, hinting at a close relationship between them (Fig. 7b). The wavelet analysis also supports the periodicities detected by REDFIT (Fig. 7b).
TOC reveals two peaks at periodicities of 119.8 and 33.2 mm (Fig. 6). Applying a Gaussian filter reveals amplitudes for the 119.8 mm periodicity that are lower at the bottom of the section but increase towards the top. The 33.2 mm periodicity exhibits generally lower amplitudes over most of the section, and only increases close to the top (Fig. 7c). Both frequencies are well represented in the wavelet analysis, although they show a slight spread. This implies that periodicities detected by REDFIT may be subject to a slight variability over time, likely caused by minor random changes in sedimentation rates (Fig. 7c).
Paleobiological proxies (
The allochthonous taxa show three peaks with periodicities of 77.61, 45.28, and 36.22 mm (Fig. 6). Filtering the periodicities reveals that the 77.61 mm periodicity exhibits quite low amplitudes in the lower part of the section, before increasing at sample 80. Following this increase the amplitude remains constant throughout the section. The two remaining periodicities show similar amplitude modulations throughout the section. They exhibit rather low amplitudes with a single excursion detected between sample 91 and 65 (Fig. 8f). The wavelet correlates well with the periodicities detected by REDFIT and shows no significant frequency drifts (Fig. 8f).
Preliminary sedimentation rate estimates can be obtained by calculating the
total mean sedimentation rate of the Laa Formation. Based on literature data,
up to 1000 m of sediment was deposited in
In the absence of any direct data on the average sedimentation rate of the
finely laminated sediments encountered in the outcrop, existing sedimentation
rate estimates from similar depositional environment in a more basin-ward
position were used as a rough baseline (Hohenegger et al., 2009; Hohenegger
and Wagreich, 2011). Cross-correlation with the 100 kyr eccentricity and
41 kyr obliquity cycles suggests an average sedimentation rate of
While this is arguably only provides a tentative indication of the actual
sedimentation rate for the Laa section, it nevertheless supports the
assumption that the cycles found within the studied section should be placed
within the temporal range of the solar band, providing a tuning target for
the encountered cycles (Solanki et al., 2004; Kern et al., 2012, 2013). This
approach, however, depends on how the frequencies of known cycles are related
to each other. A best-fit adjustment is only possible if the spectral peaks
within the data sets are spaced at similar intervals to known solar cycles.
This is similar to the principle that Milankovitch cycles always show a
relationship of approximately
While short-term variations in the sedimentation rate will naturally occur
within a natural environment, time-averaged rates of sedimentation accumulation were likely constant for the investigated section. This assumption is
supported by both REDFIT and wavelet analyses, since they show clear peaks,
with only slight evidence of peak broadening, indicative of smaller-scale
variations in sedimentation rate small enough not to mask the preserved
cycles. This indicates that sedimentation rates remained relatively constant
with only negligible variation for the studied section. Further support is
given by the presence of similar cyclic patterns in two unrelated data sets.
This relative stability of sedimentation rates can be explained by a process
of time averaging caused by the sampling distance of
Since the observed cycles show a remarkably similar spectral relationship to
known solar cycles, a best-fit adjustment can be used to tune the recorded
cycles. Considering the sedimentary history and studies in similar settings,
a sedimentation rate of 512 mm kyr
Organizing our results in this manner reveals that most detected cycles can
be roughly arranged into groups of similar periodicities: the first group
encompasses all periodicities below 65 years and thus contain three
periodicities – 60.47 (calcium carbonate), 64.84 (TOC), and 48.83 years
(
Based on these assumptions the recalculated time values now fit very well
within the reported ranges for the Upper and the Lower Gleissberg cycle, as
well as the Suess/de Vries cycle, without further need for individual
adjustment of cycles or sedimentation rates. Calculating the average of all
values that lie within the reported ranges of the cycle gives mean values of
127.22 years for all periodicities associated with the Upper Gleissberg
cycle and an average of 64.17 years for the Lower Gleissberg cycle.
Similarly, the average of periodicities associated with the Suess/de Vries
cycle now average 211.5 years (Table 3). This adjustment preserves the
Likewise, the 399.32-year periodicity fits well with a reported 400-year cycle that was detected in multiple studies of both marine and lacustrine sediments from the Holocene (Dean and Schwalb, 2000; Domack et al., 2001; Dean et al., 2002). The detected periodicity of 479.20 years (275.54 mm) may be loosely linked to an unnamed 500-year cycle (Stuiver et al., 1995; Chapman et al., 2000). Yin et al. (2007) and Kern et al. (2012) detected comparable periodicities related to this cycle in Holocene and Late Miocene records (Table 3).
The detection of cycles with higher frequencies, like the well-known
Based on the remarkably close fit of detected periodicities after tuning to
the temporal ranges of all cycles of solar variation, age estimates can be
calculated for the high-resolution section. Assuming our assumptions are
correct, the tuning results in roughly 1190 years contained within the
684 mm of the high-resolution section of Laa an der Thaya. Assuming that
time-averaged sedimentation rates did not vary significantly over the total
time represented within the 5.5 m outcrop in the clay pit of Laa an der
Thaya can be roughly estimated as no more than
Assuming that the best-fit tuning indeed reflects solar cycles, the individual periodicities of the nine different proxies can now be further evaluated with respect to how the different cycles influenced them. Significant frequencies are present in most proxies, the intensity of their response, however, differs from each other. This seems to indicate that different solar cycles are expressed in a unique way in each proxy. Kern et al. (2012, 2013) came to a similar conclusion based on their study of Late Miocene lake sediments. Further analysis of these frequencies and their relationship with regards to different proxies, however, requires at least basic information on the controlling ecological parameters. The precise factors controlling the abundance and distribution of different calcareous nannoplankton taxa, however, are still a matter of discussion, even in modern oceans (e.g., Ziveri et al., 2004). Nevertheless, some general statements can be made for key taxa that allow for interpretations concerning productivity, availability of nutrients, and surface water temperature when comparing them directly to other taxa (see Haq, 1980; Ziveri et al., 2004; Wade and Bown, 2006; Auer et al., 2014). This allows for comparison of recorded taxa in terms of relative changes to each other, focusing on their most often observed ecological preferences and describing observed cyclic patterns in the analyzed proxies. The following discussion is thus only based on the comparison of commonly referenced ecological preferences of coccolith taxa. This approach has already yielded promising results, revealing short-term small-scale changes in sea level (Auer et al., 2014).
For the purpose of this work, the ecological preferences of two taxa are of
particular importance:
For the genus
The cross-examination of the periodicities in all studied proxy data using this approach has revealed striking similarities in the periodicities for proxies that can be linked to the same environmental parameters. This coupled response of different and unrelated proxies for the same environmental parameters strongly hints at a preservation of a cyclic climate variation within these proxies linked to solar variation.
One such similarity is the strong evidence of the 119.8 mm periodicity in
TOC and
Model depicting the Suess and Gleissberg cycles and the dominant
influence they had on different paleoenvironmental systems. The (combined
upper
Comparison of these factors with a similar peak pattern in the REDFIT analysis
points towards a cyclic variation influencing primary productivity, oxygen
content, and the abundance of
A weak link between the major periodicities detected for the calcium
carbonate content and the amount of allochthonous taxa suggests that the
calcium carbonate content in the section was also influenced by the
terrigenous input into the sediment. Since coccolithophores produced more
massive coccoliths during the Cretaceous (Stanley et al., 2005) compared to
Early Miocene coccolith taxa, allochthonous taxa have a significant
contribution to the calcium carbonate content. The similarities between the
content of allochthonous taxa and calcium carbonate can thus be seen as an
expression of changing terrigenous input caused by variations in
precipitation. In addition, the same cyclicities also occur to a varying
degree in the abundance of
The fluctuations in nutrient availability also become apparent in the sulfur content, since the amount of sulfur is directly linked to lowered oxygen conditions at the sea floor caused by increased primary productivity as a result of higher fertility levels through terrigenous influx (Berner, 1981; Maynard, 1982; Berner and Raiswell, 1984). This effect may have been further amplified by increased water stratification caused by significant freshwater input, which exerted a positive feedback on the oxygen content of bottom waters, resulting in slight periodic variations in oxygen content (e.g., Schulz et al., 2005).
The influence of the similar forcing that affected different ecological
parameters also becomes apparent in a direct comparison of the filtered
amplitudes in the studied proxies with the paleoenvironmental reconstruction
based on geochemical analyses and calcareous nannoplankton assemblages of
Auer et al. (2014). A shift in amplitude occurs in some parameters that
corresponds well with the observed changes of paleoenvironmental conditions
from a more near-shore freshwater influenced setting in the lower part
compared to the more upwelling-dominated upper part of the studied section.
This shift is clearly visible in the wavelet spectra of sulfur and TOC
contents, as well as the relative abundances of
Conversely, proxies with a strong connection to precipitation, freshwater
influx, and terrigenous nutrient input, such as calcium carbonate content,
A complex link between cosmic rays emitted by the sun and the amount of cloud cover has been previously discussed (e.g., Svensmark and Friis-Christensen, 1997; Kristjánsson et al., 2004; Erlykin et al., 2010). This proposed link, while quite tenuous (see Erlykin et al., 2010), gives some evidence that solar cycles influence global climate through a complex link between variations in the cloud cover that cause variations in the planet's albedo. Similarly, there is evidence that differences in the UV irradiation of the stratosphere have a strong impact on ozone production and thus heating of the troposphere. This effect on the troposphere, in turn, may have influenced storm-tract paths and mid- to low-latitude precipitation according to Kerr (1999) and Shindell et al. (1999). The strength of the North Atlantic Oscillation in particular seems to be heavily influenced by this process on a decadal to centennial scale (Shindell et al., 2001).
This may explain how precipitation as well as wind patterns is controlled by solar variation as both are likely an expression of variations in decadal to millennial temperature patterns. Temperature variations subsequently induced the inferred variability in precipitation and wind patterns. With this considered, our results suggest that precipitation was much more sensitive to the Gleissberg cycles, whereas larger-scale wind patterns controlling upwelling appear to be linked to the longer Suess/de Vries cycle. While our results provide initial evidence of this relationship, more detailed studies are, however, needed to verify and validate this interpretation. In particular, comparable studies need to be performed in different settings to exclude system intrinsic variations in the Central Paratethys as the cause of the observed periodicities and to further test the applied best-fit tuning in sections where an independent age control is available.
The cyclic nature of climatic changes preserved in a total of nine geochemical and paleoecological proxies was studied using time series analysis (REDFIT analysis and wavelet spectra). All proxies display highly similar periodicities, although some variations in power and their exact location within the frequency band occur. Our results expand on a previous study focused on changes in paleoenvironmental conditions caused by changes in relative sea level (Auer et al., 2014) and reveal a strong influence of small-scale, short-term climate variability on local paleoenvironmental conditions caused by cyclic variations in solar activity.
With a best-fit adjustment of detected cycles to known cycles of solar
origin, the detected periodicities correspond remarkably well with cycles of
solar variation present in Holocene sun spot records and Late Miocene lake
records. Assuming the applied tuning is accurate, the section appears to be
influenced by the Lower (
Accepting the hypothesis that these cycles represent environmental responses
to known solar cycles, the integrated study of multiple paleoenvironmental
proxies revealed a complex interplay of solar variation with the environment
(Fig. 9). While variations in coccolith abundances and geochemical data show
a clear correlation with the calculated solar cycles, their response to
different cycles varies considerably. Similar responses to solar forcing were
found within proxies related to the same ecological parameters, indicating
that the observed cyclicities represent a clear influence of solar forcing on
climatic conditions. Based on this, we found that the Upper and Lower
Gleissberg cycles appear to be a driving factor for the amount of terrigenous
influx into the ocean. This link is reflected in the amount of allochthonous
taxa, a direct proxy for terrigenous input, and also the abundance of
reticulofenestrids, a proxy for freshwater influx and availability of
terrigenous nutrients. This suggests that precipitation rather than wind was
the key factor (Fig. 9). Upwelling conditions and water temperature, in turn,
were controlled by the longer Suess/de Vries cycle, reflected in a similar
response of
The authors would like to thank Stjepan Ćorić (Geological Survey of Austria, Vienna), Patrick Grunert (University of Graz), and Andrea Kern (State Museum of Natural History Stuttgart) for many helpful comments and discussions. We also would like to thank Marie-Pierre Aubry (Rutgers University) for providing key literature. Additional thanks go to the participants of the field course “Paleontological Lab- and Fieldwork” for their help with sampling and the logging of the outcrop, as well as David Strahlhofer (University of Graz) for his assistance with sample preparation in the lab. The authors would also like to thank the three anonymous reviewers for their many helpful suggestions and comments. Funding for this study was provided by the FWF (grant P-23492-B17). Edited by: A. Sluijs